[Surgeon General’s Smoking Report,  health reports Continued from Part  100]

2004 The Health Consequences of Smoking: A Report of the Surgeon General.  Office of the Surgeon General (US); Office on Smoking and Health (US). Atlanta (GA): Centers for Disease Control and Prevention (US);  Forty years of reports. Cancers of stomach, uterine cervix, pancreas, kidney, acute myeloid leukemia, pneumonia, aneurysm, cataract, periodontitis. Conclusion “smoking generally diminishes the  health of smokers.”

https://www.ncbi.nlm.nih.gov/books/NBK44698/

Part 101 The Health Consequences of Smoking: A Report of the Surgeon General. 2004. Tobacco. Diminished health status, dental, erectile dysfunction, eye disease, peptic ulcer.

6 Other Effects

Introduction

This chapter addresses evidence on smoking and health effects over a range of specific diseases and non-specific but adverse consequences. The associations reviewed appear to reflect both specific and non-specific pathways of injury by tobacco smoke. The evidence indicates that smoking should be considered not only a cause of specific diseases and conditions, but a contributing factor to nonspecific morbidity and a diminished quality of life.

Diminished Health Status

This section focuses on the question of whether cigarette smokers have poorer health in comparison with nonsmokers, beyond the already well-characterized burden of morbidity and mortality from the specific diseases caused by smoking. The hypothesis that smoking might impair health in general draws plausibility from the toxicologic richness of tobacco smoke, the well-documented systemic distribution of tobacco smoke components and metabolites, and the effects on host defenses, including the immune system. Additionally, impairment of organ function short of the level at which clinical disease is diagnosed may leave the smoker vulnerable to otherwise well-tolerated threats to health. For example, the reduction of lung function found in many smokers who do not have overt chronic obstructive pulmonary disease (COPD) may increase the risk for developing a more severe illness with a respiratory infection, or having a respiratory complication following surgery.

This section reviews studies that have addressed a number of health status indicators (Figure 6.1) including direct reports of health status or responses to an instrument that provides a health status index, and indirect indicators such as medical services utilization data. When interpreting the findings of these studies, consideration needs to be given to the potential causal pathways linking smoking to a poor health status, the assessment and measurement of health status, and the potential for biases, such as from confounding, to affect associations of smoking with these outcome measures.

Figure 6.1

A conceptual model for the relationship between cigarette smoking and diminished health status. 

For the diseases caused by smoking, direct causal pathways are implicit. For example, substantial evidence supports the hypothesis that smoking causes lung cancer through the direct deposition of tobacco smoke carcinogens in the respiratory tract. For some of the outcome measures considered in this section, pathways are far less certain and may be both direct and indirect. Increased absenteeism might reflect, for example, the tendency of smokers to have more severe respiratory illnesses than nonsmokers, possibly attributable to the effects of smoking on respiratory defenses or because smokers tend to have a lower level of lung function.

The outcomes considered in this section have multiple determinants. Health status itself is an integrative measure reflecting the net consequences of the many varied factors that determine health and well-being. To the extent that smokers differ from non-smokers in these factors, there is a potential for confounding to distort associations of smoking with the outcome measures. Studies show, for example, that smokers and nonsmokers differ in aspects of lifestyle and in their approaches to health care (e.g., the use of preventive services such as multiphasic testing [Oakes et al. 1974] and screening [Beaulieu et al. 1996; Edwards and Boulet 1997]). Additionally, the suite of relevant confounding factors may differ from outcome to outcome, and for some outcomes there is uncertainty as to the relevant confounding factors. Some of the individual characteristics that affect the decision to start smoking and to continue to smoke also may be determinants of risk for the outcomes considered here.

Conclusions of Previous Surgeon General’s Reports

Extensive research over time has identified cigarette smoking as a cause of specific diseases, and many reports from the Surgeon General have focused on smoking and these diseases. These reports have also addressed more general and nonspecific adverse consequences of smoking, such as increased rates of absenteeism from work or the utilization of medical services among smokers in comparison with nonsmokers. Conclusions from the reports that relate to these outcomes are listed in Table 6.1, including findings on general respiratory morbidity. Reports of increased morbidity from common and frequent viral and bacterial respiratory infections among smokers have been reviewed (U.S. Department of Health and Human Services [USDHHS] 1990) and are among the topics covered in Chapter 4 of this report. However, the overall health status of smokers compared with nonsmokers has not been comprehensively addressed in prior Surgeon General’s reports.

Table 6.1

Conclusions from previous Surgeon General’s reports concerning smoking as a cause of diminished health status and respiratory morbidity. 

Biologic Basis

Cigarette smoke, inhaled through the mouth into the lungs, reaches lung airways and alveoli, where the tobacco smoke components pass into the systemic circulation (Murray 1986). The airways and alveoli themselves are exposed to the gaseous and particulate components of tobacco smoke as many of these components readily pass through the alveolar-capillary membrane into the alveolar capillaries and then circulate throughout the body. Nicotine, for example, which is among these components, reaches the brain within 10 seconds after smoke is inhaled (USDHHS 1988). It is distributed throughout the body and has been found in breast milk (Schwartz-Bickenbach et al. 1987; Schulte-Hobein et al. 1992; Golding 1997) and in cervical mucus (Prokopczyk et al. 1997). Carbon monoxide, a diffusible gas, moves from the alveoli into the capillaries where it binds tightly to the hemoglobin of the red blood cells. Benzo[a]pyrene, a well-characterized carcinogen in tobacco smoke, can be found bound to the blood cells in the epithelial cells of the airways of smokers and in their major organs. The effects of smoking on host defenses and aspects of immune function have been covered in prior reports (USDHHS 1990, 1994) and again in this report. These effects may have the consequence of increasing risks for infections, whether of the respiratory tract or other organs. However, there has been less research to date on infections beyond those of the respiratory tract. This systemic distribution of tobacco smoke components underlies the associations between smoking and disease that are well documented for many organs including cardiovascular disease, stroke, and cancers of the kidney and urinary bladder. The widespread distribution may also lead to more general effects on health.

This same systemic distribution may have non-specific effects as well, contributing to a reduction in health status. Exposure to tobacco smoke components causes smoke-specific diseases such as bladder cancer (carcinogens in urine come in contact with the bladder) and atherosclerosis, probably reflecting multiple underlying mechanisms with inflammation having a central role (Cross et al. 1999). Underlying mechanisms might include heightened oxidative stress and reduced antioxidant defenses, increased inflammatory activity, reduced host defenses against infection, and lowered reparative capacities of tissues. The evidence on these mechanisms is at varying levels of development. This section focuses on oxidative stress as an example, selected because the available literature is extensive.

Oxidative Stress

Oxidative stress refers to an increased exposure to oxidants and/or a decreased antioxidant capacity, caused by oxygen radicals that mutate DNA, promote atherosclerosis, and lead to chronic lung injury. Oxidative stress is now hypothesized to be a general mechanism underlying aging and many of the chronic diseases associated with aging, contributing to the development of cancer, cardiovascular disease, and COPD (Ames et al. 1995). Mounting evidence points to chronic oxidative stress as one mechanism whereby smoking affects health. Smoking is associated with evidence of chronic systemic inflammation, perhaps a consequence of the chronic oxidative stress experienced by the smoker (Cross et al. 1999; Hecht 1999). The oxidant load posed by cigarette smoke is substantial; the tar component is estimated to contain 1018 oxygen radicals per gram of tar and the gas component to have as many as 1015 other organic radicals per puff (Repine et al. 1997).

A number of comparisons between smokers and nonsmokers have been made with respect to measures of biomolecular oxidative damage, including oxidative injury to DNA, proteins, and lipids. A widely used assay for quantifying oxidative damage to DNA is 8-hydroxydeoxyguanosine (8-OH-dG). The assay measures hydroxyl radical-induced DNA damage at C8 of guanine (Lagorio et al. 1994), which has been linked experimentally to cigarette smoke condensate (Leanderson and Tagesson 1990). Cultured human lung cells exposed to cigarette smoke had 70 percent higher 8-OH-dG levels than unexposed cells (Leanderson and Tagesson 1992). DNA from the lung tissue of smokers had 42 percent higher 8-OH-dG levels than the DNA from nonsmokers, and 8-OH-dG concentrations increased according to the number of cigarettes smoked per day (Asami et al. 1997).

Studies comparing 8-OH-dG levels in DNA from smokers and nonsmokers are summarized in Table 6.2. In general, regardless of the biologic material, smokers tend to have greater damage. A strong dose-response association with the number of cigarettes smoked was observed in one study (Lodovici et al. 2000), but an inverse dose-response trend was observed in another (van Zeeland et al. 1999). When levels of 8-OH-dG in circulating lymphocytes were compared before and after cigarettes were smoked, Kiyosawa and colleagues (1990) observed that 8-OH-dG levels increased 54 percent after smoking. A similar but less frequently used approach to determine biomolecular oxidative damage is to assay 8-hydroxyguanine, which has been found in leukocyte DNA (Asami et al. 1997) and in urine (Suzuki et al. 1995) of smokers at concentrations at least 90 percent higher than in nonsmokers.

Table 6.2

Studies on the association between smoking and oxidative injury. 

Oxidative damage to proteins can occur in both amino acid residues and the peptide backbone in protein, and can be assessed by assaying protein carbonyls (Reznick et al. 1992; Eiserich et al. 1995). Studies document that exposing human plasma (Reznick et al. 1992; Eiserich et al. 1995; Panda et al. 1999) or saliva (Nagler et al. 2000) to cigarette smoke increased protein carbonyl concentrations by more than 300 percent. Compared with unexposed guinea pigs, guinea pigs exposed to cigarette smoke had plasma protein carbonyl concentrations more than 30 times greater (Panda et al. 2000). In humans, protein carbonyl concentrations in 15 smokers were 61 percent higher than in 5 comparison nonsmokers (Lee et al. 1998).

Isoprostanes constitute a specific measure of lipid peroxidation and serve as good general markers of oxidative injury (Morrow and Roberts 1996). Free radicals catalyze the peroxidation of arachidonic acid to F2-isoprostanes (Morrow and Roberts 1996). Circulating (Morrow et al. 1995) and urinary (Morrow et al. 1995; Reilly et al. 1996) isoprostane levels have been shown to be markedly higher in smokers than in non-smokers (Table 6.2). Circulating (Morrow et al. 1995; Pilz et al. 2000) and urinary (Reilly et al. 1996; Pilz et al. 2000) isoprostane concentrations decreased at least 20 percent within two weeks of smoking cessation. Babies of smoking mothers had concentrations of isoprostane levels in their umbilical arteries and veins more than 110 percent higher than babies of nonsmoking mothers (Obwegeser et al. 1999).

Another widely used measure of free radical catalyzed lipid peroxidation is thiobarbituric acid reactive substances (TBARS) (Bonithon-Kopp et al. 1997). Comparisons of TBARS between smokers and nonsmokers have shown that (1) current smokers have higher TBARS levels—sometimes strikingly higher, (2) levels of TBARS rise after smoking, and (3) the influence of smoking on increased lipid peroxidation can be offset somewhat by administering the antioxidant micronutrients vitamins C and E (Table 6.2).

Antioxidant Depletion

Even as smokers are exposed to the oxidative stress of regularly inhaling cigarette smoke, substantial evidence shows that blood levels of individual antioxidant micronutrients are lower in current smokers than in nonsmokers. This association has been clearly demonstrated for vitamin C (McClean et al. 1976; Bolton-Smith et al. 1991; Ross et al. 1995; Lykkesfeldt et al. 1997) and for total and selected carotenoids including ?-carotene, ?-carotene, and cryptoxanthin (Aoki et al. 1987; Stryker et al. 1988; Bolton-Smith et al. 1991; Pamuk et al. 1994; Ross et al. 1995; Brady et al. 1996; Alberg et al. 2000). For vitamin C (Brook and Grimshaw 1968; Buiatti et al. 1996; Marangon et al. 1998) and several of the specific carotenoids (Comstock et al. 1988; Nierenberg et al. 1989; Buiatti et al. 1996; Marangon et al. 1998), circulating concentrations tend to decline with increasing number of cigarettes smoked.

Whether the differences in antioxidant levels across smoking categories reflect direct depletion or differing dietary intake has been controversial. If smoking directly depletes antioxidant micronutrients, the effect would presumably be acute. In fact, levels of vitamin C and selected carotenoids increased when measured in persons after 84 hours without smoking a cigarette (Brown 1996), and an experimental exposure of plasma equivalent to six puffs of cigarette smoke completely depleted the ascorbic acid present in the serum (Handelman et al. 1991; Eiserich et al. 1995). When measurements were taken at baseline and 20 minutes after smoking a cigarette, decreases in circulating micronutrient concentrations were observed (Yeung 1976).

Smoking and the Leukocyte Count

Studies show that smokers when compared with nonsmokers have generally heightened inflammation, increased white blood cell counts that remain elevated after cessation, and increased levels of other markers of inflammation such as C-reactive protein (Allen et al. 1985; Das 1985; de Maat et al. 1996; Tracy et al. 1997; Danesh et al. 1999).

The association between smoking and the leukocyte count has been extensively investigated, with numerous studies showing that current smokers have higher leukocyte counts than nonsmokers (Table 6.3). In most studies, the increase was 20 percent or more in smokers compared with nonsmokers and was present across strata of age, gender, and race. The leukocyte count increases with the number of cigarettes smoked per day and with the depth of inhalation. Similar dose-response trends were evident in other studies that did not lend themselves to inclusion in the summary tables (Petitti and Kipp 1986; Schwartz and Weiss 1991). Dose-response trends tend to be weaker when examined in relation to either pack-years1 or duration of smoking, suggesting that smoking has an immediate effect on the leukocyte count.

Table 6.3

Studies on the association between current smoking and white blood cell counts. 

The findings from former smokers are consistent with both an immediate and a persistent effect of smoking. In comparisons with lifetime nonsmokers (Table 6.4), former smokers consistently have higher white blood cell counts, but the difference is smaller than that between current smokers and lifetime nonsmokers. In most of the studies, the leukocyte counts for former smokers were only about 5 percent greater than those for lifetime nonsmokers. The excess is persistent (Petitti and Kipp 1986; Schwartz and Weiss 1991; Sunyer et al. 1996), although it decreases with increasing duration of cessation, becoming closer to the average counts found in lifetime nonsmokers (Yarnell et al. 1987; Hansen et al. 1990b). A short-term (overnight) abstention from cigarettes did not strongly influence the counts (Noble and Penny 1975).

Table 6.4

Studies on the association between former smoking and white blood cell counts. 

Prospective cohort studies have tracked changes in leukocyte counts in relation to changes in smoking. In a study of Kaiser Permanente enrollees in the San Francisco Bay area, the leukocyte counts increased 12 percent among those who started smoking during the follow-up, but it decreased 7 percent among smokers who had quit during the follow-up (Friedman et al. 1973). In a subsequent study that compared leukocyte counts of 9,392 persistent smokers with those of 3,825 smokers who had quit, the quitters experienced significantly higher declines (Friedman and Siegelaub 1980). In a cohort of homosexual men seronegative for human immunodeficiency virus (HIV), Sunyer and colleagues (1996) observed that decreases in smoking were followed by decreased white blood cell counts, and increases in smoking were followed by increased white blood cell counts. Furthermore, changes in white blood cell counts were proportional to changes in smoking patterns (Table 6.5).

Table 6.5

Studies on the percentage difference in white blood cell counts stratified by smoking patterns. 

These observations of inflammatory markers, particularly the leukocyte counts, are consistent with the induction of systemic chronic inflammation in smokers, perhaps reflecting the substantial oxidant load from habitual cigarette smoking. Studies of former smokers suggest that this state of inflammation does not simply reflect an acute effect. These observations support one of the mechanisms, oxidative stress, proposed as contributing to the general effects of smoking on health.

Epidemiologic Evidence

Absenteeism

Absenteeism from work is frequent and costly (Steers and Rhodes 1978); its multiple causes include individual and organizational factors (Steers and Rhodes 1978). Researchers investigating the effect of smoking on absenteeism face the challenges of controlling for potential confounding by individual-level factors such as alcoholism, and specifying how smoking could act in combination with other factors at both individual and group levels. While the literature is extensive (Table 6.6), the studies vary in the success with which these challenges have been met, partially reflecting the extent and quality of available data.

Table 6.6

Studies on the association between current smoking and absenteeism. 

Current Smokers

In studies with varying designs conducted in diverse locations, cigarette smokers consistently have had higher rates of absenteeism than nonsmokers (Table 6.6). The evidence also indicates that the duration of sickness absences tends to be longer for smokers and smokers miss more cumulative worktime than nonsmokers. The association between smoking and absenteeism has been observed in both men and women of all ages. Sickness absences have been measured in a variety of ways, including lost worktime per unit of time, episodes of absenteeism, and the duration of absences. The finding that smoking is associated with absenteeism, regardless of the index used, documents consistency of the observed association. Although most studies were cross-sectional or retrospective in design, two were prospective cohort studies (North et al. 1993; Niedhammer et al. 1998) and another studied smoking histories in relation to work-place attendance records during the preceding nine years (Holcomb and Meigs 1972). The findings of these prospective studies confirm that smoking preceded the absenteeism. In a few studies, the association with smoking was observed primarily in men but not in women (Green et al. 1992; North et al. 1993), but in general the findings have been consistent across all of the subgroups studied. Of the 30 studies that were the sources for the data abstracted into Table 6.6, 17 studies found that absenteeism among smokers was at least 20 percent greater than among nonsmokers in all subgroups.

Two additional reports not included in the table also provide evidence of an association between smoking and absence frequency (Ferguson 1973; Donaldson et al. 1999). In a study of 516 men employed in four occupational groups in Australia, Ferguson noted that “. . .the employee with repeated absence also tended (p <0.10), more often than the resister” (employee without repeated absences) “. . .to smoke more than 15 cigarettes daily” (Ferguson 1973, p. 336). In a study of 146 lumber company employees, a tobacco use scale was not correlated (r = 0.01) with absenteeism (Donaldson et al. 1999).

In several studies summarized in Table 6.6 that assessed the relationship between current smoking and absenteeism (Athanasou 1979; Andersson and Malmgren 1986; Hawker and Holtby 1988; Bertera 1991), current smokers were compared with all nonsmokers, including former smokers. As discussed in the following section, absenteeism rates among former smokers are persistently elevated compared with those of lifetime nonsmokers. Thus, using an “unexposed” comparison category that includes former smokers along with lifetime nonsmokers will dilute associations that would be estimated when using a “pure” unexposed category consisting solely of persons who have never smoked.

In the two studies that assessed the dose-response relationship with the number of cigarettes smoked, the likelihood of being absent increased strongly with the number of cigarettes smoked per day (Lowe 1960; Holcomb and Meigs 1972). In a retrospective cohort study of 226 male factory employees in Connecticut that included eight years of follow-up, the rate of long-term absences increased 43 percent, 57 percent, and 100 percent compared with nonsmokers for those who smoked less than one pack, one pack, and more than one pack of cigarettes per day, respectively (Holcomb and Meigs 1972). In a study of more than 3,300 male General Electric employees in England, the number of days absent for medical reasons increased 11 percent, 13 percent, 26 percent, and 57 percent compared with nonsmokers for those who smoked 1 to 9, 10 to 19, 20 to 29, and 30 or more cigarettes per day, respectively (Lowe 1960).

This body of evidence shows increased absenteeism among smokers, while providing only limited information on the reasons for the absences. A significant proportion of sickness absences in smokers would be expected to be due to smoking-associated illnesses. Athanasou and colleagues (1981)hypothesized that smoking acts as a susceptibility factor, increasing the risks for other harmful occupational exposures. In one study, smoking was associated with a significantly increased likelihood of absences resulting from problems as diverse as back symptoms, digestive tract symptoms, and neck and upper limb symptoms (Dimberg et al. 1989). A recent review summarizing 38 studies showed an increased risk for back pain in smokers compared with nonsmokers in the majority of studies (Goldberg et al. 2000). In another study, absences were elevated not only for “medical reasons” but also for “other” reasons (Lowe 1960). Substantial evidence also documents that smokers are more likely than non-smokers to have on-the-job injuries (Lowe 1960; Naus et al. 1966; Reynolds et al. 1994; Forrester et al. 1996). Because smoking increases absences for a broad set of health problems, and not just specific smoking-associated illnesses, the underlying causal pathways are likely to be multiple and general, reflecting the systemic nature of the effects of smoking.

Former Smokers

The evidence is consistent that former smokers are less likely to be absent from work compared with persistent smokers. Former smokers tend to have somewhat higher absenteeism rates than persons who have never smoked (Table 6.7), but the increases are much smaller than those for current smokers. The analyses performed by Wooden and Bush (1995) with former smokers (n= 4,812) in the 1989–1990 Australian National Health Survey illustrate the seemingly paradoxic relationship between quitting smoking and absenteeism. In a multiple regression model that included both the duration of active smoking and time since quitting, the number of years that a former smoker had smoked remained a strong predictor of absenteeism, and the likelihood of absences declined gradually over time since cessation (Wooden and Bush 1995). Similarly, Manning and colleagues (1989) found differences between recent and sustained quitters, and observed considerably higher absenteeism rates for recent quitters compared with long-term quitters. These results indicate that both prior smoking history and time since quitting are factors strongly associated with absenteeism, but in opposite directions. This pattern may arise because some smokers may quit when diagnosed with an illness caused by smoking, and the recent quitters may thus already have a smoking-induced illness that predisposes them to lost worktime.

Table 6.7

Studies on the association between former smoking and absenteeism. 

In interpreting evidence linking smoking to a diminished health status, including absenteeism, untangling the direct effects of smoking from the indirect effects is challenging, as smokers and nonsmokers may differ in potential confounding factors. Nonetheless, given the scope of the evidence available and the diversity of the populations studied, the literature does provide insights into the role of smoking as a cause of absenteeism.

With regard to confounding, alcohol use is a major factor of concern. Alcohol use has been linked to absenteeism in some studies, and smokers drink more than nonsmokers (Smith 1970; Turner 1988; Ault et al. 1991; Marmot et al. 1993; Vasse et al. 1998). Smokers are also more likely to be heavy alcohol drinkers and to use illicit substances (Merrill et al. 1999; Best et al. 2000; Brain et al. 2000; Dawson 2000), and heavy alcohol and illicit substance use, rather than cigarette smoking, could increase the likelihood of workplace absences. Studies that adjusted for alcohol consumption have generally (Hendrix and Taylor 1987; Bush and Wooden 1995; Wooden and Bush 1995), but not universally (Ault et al. 1991), found smoking to be associated with frequent absences, implying that the association of smoking with alcoholism is not due to confounding. Studies were not found that accounted for illicit substance use in assessing the association between smoking and workplace absences. Less likely is the possibility that the association between smoking and absences reflects confounding by characteristics that are linked both to smoking (see the section on “Health Status” later in this section) and to an increased risk for frequent absences. For example, women are consistently absent from work more often than men (Leigh 1983; Pines et al. 1985; Steinhardt et al. 1991). But women assume a disproportionate share of family responsibilities such as staying home with sick children, and the relative importance of smoking may therefore be less. Observations of persons with “psychosocial problems” (Leijon and Mikaelsson 1984) and anxiety/neuroses (Taylor 1968; Ferguson 1973) document increased risks for absenteeism, and if such persons are more likely to smoke, confounding is possible. Given the range of populations studied, confounding by psychosocial factors seems unlikely.

Of the relevant pathway factors leading to health-related absences, age is the primary demographic characteristic that is a potential modifying or confounding factor. Socioeconomic status, another potential confounding or modifying factor, is inherently restricted in studies within occupational groups. Age is associated with both absenteeism (Pines et al. 1985) and health status. The association between smoking and absenteeism has been observed consistently across a broad spectrum of age strata in the summarized results, implying that the association does not reflect confounding by age.

Only a few studies provide prospective data concerning absenteeism following smoking cessation; the findings suggest that smoking cessation is associated with better attendance at work. A particularly informative study conducted with employees of a North Carolina pharmaceutical company compared the attendance patterns of former smokers before and after quitting with attendance patterns of a matched group of persistent smokers (Jackson et al. 1989). In the time preceding smoking cessation by the cessation group, the persistent smokers tended to have fewer absences than the smokers who went on to stop smoking. However, during the three years following cessation, the mean number of annual sick days declined among those who quit. Absences continued to increase for persistent smokers, leading to a widening gap in absences between the two groups. The study was small, with only 70 persons participating. In a randomized trial of nine worksite smoking cessation programs, employees who were smokers at baseline had a significant reduction (p = 0.002) in self-reported sick days after stopping smoking (Jeffrey et al. 1993). In another study evaluating a workplace health promotion program that reduced smoking prevalence, the authors reported significant reductions in absenteeism for program participants but not for nonparticipants (Wood et al. 1989).

The evidence that reduced absenteeism follows cessation complements findings based on comparisons of current smokers with nonsmokers. The reduced rate after cessation supports a causal interpretation, rather than attributing the association to an indirect pathway or to confounding factors.

In summary, there is consistent evidence demonstrating that employees who are current smokers have a greater likelihood of absences from work compared with employees who have never smoked. Additional evidence is needed on dose-response trends and, more importantly, on changes in absence rates before and after smoking cessation. Other reviewers have concluded that reduced absenteeism could lead to potential savings that can be accrued from smoking cessation programs in the workplace (Kristein 1983; Warner et al. 1996).

Medical Services Utilization

Medical services utilization provides another measure of the global effects of smoking on health. The most important utilization indicators in studies on smoking can be grouped into three general categories: (1) costs, (2) outpatient visit rates, and (3) hospitalization rates. Interpreting these findings requires consideration of the many factors influencing medical services utilization. Smokers, for example, are less likely than nonsmokers to use preventive services such as screening (Beaulieu et al. 1996; Edwards and Boulet 1997). However, the high incidence of smoking-induced diseases among smokers will tend to drive their medical care needs. The socioeconomic and educational differences between smokers and nonsmokers also complicate data interpretation because of potential confounding. Comparisons of smokers within well-defined groups, such as particular workforces or health care plans, should provide unbiased comparisons.

Costs

In evaluating the relationship between smoking and medical care costs, only those studies directly addressing expenditures were considered (Table 6.8). The literature on comparative lifetime costs of medical care for smokers and nonsmokers based on assumed models and projections was not considered relevant to this chapter. Of the seven studies reviewed, six showed the medical costs of smokers to be greater by at least 15 percent in at least one subgroup. In one study of enrollees in a health maintenance organization, smokers had costs 25 percent higher than nonsmokers among those younger than 65 years of age, but few differences were observed in those age 65 years or older (Terry et al. 1998). Only the study by Vogt and Schweitzer (1985) on enrollees in Kaiser Permanente found no differences between smokers and nonsmokers.

Table 6.8

Studies on the association between current smoking and medical service costs. 

Two studies not included in Table 6.8 are also relevant. In a population of retirees followed for one year, smoking was associated with added health care costs of more than $1,900 per year per pack of cigarettes smoked per day, after adjusting for age, gender, education, seat belt use, and alcohol consumption (Leigh and Fries 1992). In a study conducted as part of a worksite health promotion program in Birmingham, Alabama, smokers were found to have incurred more costs than nonsmokers, but the data were not presented (Weaver et al. 1998).

Outpatient Services

In several studies (Table 6.8), smokers were at least 15 percent more likely than nonsmokers to use outpatient services (Peters and Ferris 1967; Palmore 1970; Chetwynd and Rayner 1986; Freeborn et al. 1990); one study found an increased likelihood of 6 percent (Rice et al. 1986). In studies that stratified age and gender, strong associations with smoking were observed in particular groups. Male smokers were more frequent users of outpatient services than were male nonsmokers, but this difference was not found among females in one study (Oakes et al. 1974). In another study, this gender difference occurred in young but not old persons (Ashford 1973). Three studies showed only small differences in the use of outpatient services between smokers and nonsmokers (Vogt and Schweitzer 1985; Halpern and Warner 1994; Miller et al. 1999).

The frequency of outpatient visits does not appear to increase with the number of cigarettes smoked (Peters and Ferris 1967; Balarajan et al. 1985; Marsden et al. 1988). However, regardless of the number of cigarettes smoked, some studies documented a large difference in the number of visits by smokers compared with nonsmokers.

Hospitalization

In all but one of the studies considered (Terry et al. 1998), smokers had higher hospitalization rates than nonsmokers; the differences were at least 10 percent. In two other studies that stratified age and gender, one study found an association in males but not in females (Oakes et al. 1974), and the other study found an association only among younger females (Ashford 1973).

Additional studies corroborate the results summarized in Table 6.8. In a study of a cohort of retirees followed for one year, the number of packs of cigarettes smoked per day was significantly associated with the number of days hospitalized (Leigh and Fries 1992). In a study of 1,000 veterans accessing the Veterans Administration system in Connecticut, tobacco users were significantly more likely (p <0.01) than nonusers to be hospitalized, and tobacco users were significantly more likely (p<0.01) than nonusers to be hospitalized and to spend more days in the hospital (Benedetto et al. 1998). In a study of Kaiser Permanente enrollees in Oregon, Pope (1982) observed a weak, non-significant correlation between a smoking index and hospitalization rates in the youngest age group for men and women (aged <35 years), but this association was not present in the other age groups studied.

Dose-response data are available from two prospective cohort studies (Table 6.9). In the Coronary Drug Project, the five-year hospitalization rates for smokers compared with nonsmokers plateaued at the lowest smoking category, and were more compatible with a threshold relationship than with a nonthreshold dose-response relationship. However, it was unclear whether these analyses accounted for the higher mortality rates experienced by smokers relative to nonsmokers during the follow-up period (Coronary Drug Project Research Group 1976). In a two-year follow-up of smokers in the American Cancer Society Cancer Prevention Study I (CPS-I) a strong dose-response relationship was present: compared with those who smoked 1 to 9 cigarettes per day, those who smoked 10 to 19, 20 to 39, and 40 or more cigarettes per day had an increased likelihood of hospitalization during the follow-up period of 8.5 percent, 14.6 percent, and 28.0 percent, respectively (Hammond 1965). In a cross-sectional survey of U.S. military personnel that compared smokers with nonsmokers, those who smoked one-half of a pack or less, one pack, and one and one-half packs or more per day had increases in self-reported days hospitalized of 28.1 percent, 6.3 percent, and 54.7 percent, respectively (Marsden et al. 1988).

Table 6.9

Studies on the association between the amount smoked and medical service utilization rates. 

Former Smokers

Studies comparing the use of medical services by former smokers with lifetime nonsmokers are summarized in Table 6.10. Costs were 26 percent higher for former smokers in one study (Pronk et al. 1999), and higher for some services but not higher overall in another study (Vogt and Schweitzer 1985). In every study, former smokers were more likely than lifetime nonsmokers to use outpatient services. In a study conducted in the United Kingdom that was stratified by age and gender, smokers were more likely than non-smokers to have general practice health care providers visit their homes for an illness (Ashford 1973). The use of outpatient services by smokers remained elevated compared with that of nonsmokers long after smoking cessation (Halpern and Warner 1994). For hospitalizations the findings were mixed, with three studies showing higher rates in former smokers (Van Peenen et al. 1986; Kaplan et al. 1992; Halpern and Warner 1994). In one of these studies, however, the difference was eliminated after adjusting for age, and in two other studies there were only small differences between former smokers and lifetime nonsmokers. In another study that stratified age and gender, former smokers were more likely than lifetime nonsmokers to be hospitalized in some strata, but less likely in others, without a consistent pattern (Ashford 1973).

Table 6.10

Studies on the association between former smoking and medical services utilization costs and rates. 

These studies generally have not taken into account prior smoking history and time since quitting, nor have they considered whether development of a disease led to quitting. The extent of smoking before quitting is a determinant of risk, and risks fall for many diseases as the duration of quitting lengthens. The somewhat inconsistent findings may reflect (1) the heterogeneity of former smokers in these studies and (2) analysis strategies that did not fully account for risk determinants in the former smokers. In an analysis of the 1990 National Health Interview Survey data that accounted for time since quitting, former smokers had significantly more hospital admissions until 10 years following cessation, at which point former smokers and lifetime nonsmokers had similar numbers of hospital admissions (Halpern and Warner 1994).

The clinical trials of Wagner and colleagues (1995) provide additional evidence. Two cessation trials followed participants and collected medical care utilization data. After six years of follow-up, quitters experienced reductions in outpatient visits, hospital admissions, and hospital days in both trials compared with persistent smokers. In contrast, medical care utilization continued to increase among persistent smokers: 7 to 15 percent for outpatient visits, 30 to 45 percent for hospital admissions, and 75 to 100 percent for days spent in the hospital. These divergent patterns in the use of medical care services resulted in substantially greater rates of hospitalization, hospital days, and outpatient visits for persistent smokers.

Age

Several studies suggest that smoking may have a greater impact on the youngest age groups compared with older age groups. More frequent use of outpatient (Peters and Ferris 1967; Newcomb and Bentler 1987) and inpatient (Newcomb and Bentler 1987) services among smokers than among nonsmokers has been observed even in adolescents and young adults, suggesting that the differences observed in smoking and nonsmoking older adults are not solely a result of smoking-induced diseases. In fact, in a few studies higher levels of service utilization were observed among smokers than among nonsmokers in the younger age groups, but such differences were either not present or were reversed in the oldest age groups. This pattern is evident in the cross-sectional analyses of the 1970 U.S. National Health Interview Survey data, a random sample of U.S. households in which both smoking men and smoking women had a markedly higher number of days hospitalized per year than their nonsmoking counterparts until they reached their mid-40s, at which point the differences between smokers and nonsmokers became more subtle (Weinkam et al. 1987).

In general, compared with nonsmokers, smokers tend to incur more medical costs, to see physicians more often in the outpatient setting, and to be admitted to the hospital more often. Among patients admitted to the hospital, smokers have longer lengths of stay and incur greater expenses per admission than nonsmokers. Less information is available concerning the use of medical services such as prescription drugs and emergency department visits, but increases for smokers compared with nonsmokers have also been observed with respect to these outcomes (Chetwynd and Rayner 1986; Miller et al. 1999). Although smokers use more palliative care services, as demonstrated by this review, smokers have been less likely than non-smokers to use preventive services such as multiphasic testing (Oakes et al. 1974) and screening (Beaulieu et al. 1996; Edwards and Boulet 1997).

Postoperative Complications

In comparison with nonsmokers, smokers have been hypothesized to be at a higher risk for postoperative complications because of a greater frequency of chronic diseases, impaired pulmonary reserve, altered immune responses, and impaired wound healing. Higher rates of postoperative complications in smokers could contribute to the greater costs that they incur for health care services.

Substantial clinical and experimental research has been conducted on the relevant effects of smoking on host defenses, immune responses, and wound healing. As reviewed elsewhere in this report and in a previous Surgeon General’s report (USDHHS 1990), smoking produces a range of effects on respiratory defense mechanisms that may increase the risk for postoperative pneumonia. Compromised lung function and the presence of COPD increase the risks for respiratory complications, including respiratory failure. The increased likelihood of coronary heart disease (CHD) in smokers increases the risk for cardiac events during and after surgery. In animal and clinical models, exposure to tobacco smoke and nicotine specifically impaired aspects of wound healing (Brown et al. 1986; Silcox et al. 1995; Haverstock and Mandracchia 1998; Jorgensen et al. 1998; Hollinger et al. 1999).

The literature on postoperative complications is extensive and diverse in the scope of complications associated with smoking. Table 6.11 provides evidence for lower survival rates after surgery for smokers compared with nonsmokers and suggests that this increased mortality may reflect a range of specific and nonspecific consequences of smoking, including a greater risk for postoperative complications related to the surgery. A number of reports address specific surgical complications such as flap failures, wound infections, and poor orthopedic outcomes. A similarly diverse set of reports consistently shows that smoking also increases the risk of respiratory complications.

Table 6.11

Studies on the association between smoking and complications of surgery. 

Health Status

Comparisons of self-rated health statuses in smokers and nonsmokers provide further evidence of the global effects of smoking on health. Although self-ratings are inherently subjective, they provide direct evidence of the relationship of smoking to a diminished health status. Consonant with the complex concept of “health,” health status is itself a multidimensional construct, challenging to measure and approached with varied measurement methods, including direct questions on perceived health status and standardized scales. For example, the Short Form 36 (SF-36) is a standardized, 36-item scale that measures eight dimensions of health (Lyons et al. 1994), three of which have a direct relevance to this review: general health perceptions (five items), physical health (four items), and mental health (five items). Table 6.12 (smokers versus nonsmokers), Table 6.13 (dose-responses), and Table 6.14 (former smokers versus non-smokers) summarize the evidence. Studies were grouped according to the aspect of health status measured: symptoms/illnesses/health complaints, perceived health status (poor/good), physical function, physical status, general health status, life satisfaction/ dissatisfaction, well-being, quality of life, tiredness, and mental health. In some studies “poor” health was measured whereas in others “good” health was measured, so the anticipated directions of the effects of smoking vary with the specified outcome.

Table 6.12

Studies comparing the health status of smokers and nonsmokers. 

Table 6.13

Studies evaluating the dose-response relationship between the number of cigarettes smoked per day and health status. 

Table 6.14

Studies comparing the health status of former smokers and nonsmokers. 

Studies with varying designs, as well as studies measuring physical health status (Table 6.12), have shown uniformly that smokers tend to rate their general health status lower than do nonsmokers. Studies that do not include sufficient data to summarize in the tables obtained similar results. A study of 558 Bank of America retirees in California comparing smokers with nonsmokers showed that smoking was strongly associated with a higher number of sick days confined to home (Leigh and Fries 1992). In an analysis of 1990 National Health Interview Survey data, the perception of health status held by current smokers was significantly lower than that held by nonsmokers (Erickson 1998). In a multiple regression analysis of data collected from approximately 18,000 men and women in Finland, which included variables for sociodemographic characteristics, family life, morbid conditions, pain, psychosocial problems, and relative weight, smoking was associated with a significantly lower perceived health status in men but not in women (Fylkesnes and Førde 1991). In a random sample of 1,200 adults in South Wales, United Kingdom, the mean score on the SF-36 general health perception scale among participants who had ever smoked was 7.8 points lower than for those who had never smoked (Lyons et al. 1994). A study using the same scale with 921 U.S. male military veterans showed that current smoking was significantly inversely correlated with good general health perceptions (Schnurr and Spiro 1999). In a telephone survey of Newfoundland residents, the likelihood of rating one’s health as good declined in proportion to the number of cigarettes smoked per day; those who had never smoked were more than four times more likely than smokers of more than 30 cigarettes per day to rate their health as good (Segovia et al. 1989). In a survey of 1,623 patients from nine medical practices in Scotland who had a history of smoking, persistent smokers rated their general health 8.0 percent lower than former smokers rated theirs on the SF-36 scale (Tillmann and Silcock 1997). Among 2,502 enrollees in an Oregon health maintenance organization, smoking was negatively correlated with general health status for both men and women, an observation that extended to measures of mental and physical health status (Pope 1982).

Smokers in at least one subgroup were at least 10 percent more likely than nonsmokers to rate their health as poor, including studies that compared self-reported chronic conditions (Balarajan et al. 1985; Halpern and Warner 1994), acute conditions (Balarajan et al. 1985), and physical symptoms (Macnee 1991; York and Hirsh 1995). An increasing number of cigarettes smoked per day was consistently associated with increased risks for symptoms or illnesses (Balarajan et al. 1985; Marsden et al. 1988; Joung et al. 1995), and with a greater likelihood of rating one’s health as poor (Joung et al. 1995; Poikolainen et al. 1996; Manderbacka et al. 1999) (Table 6.13), with differences between the highest and lowest exposure categories of about 30 percent or greater in every study that assessed dose-response trends (Table 6.13). For several measures of poor health, the differences between former smokers and lifetime nonsmokers (Table 6.14) tended to be even more striking than for comparisons between current smokers and lifetime nonsmokers, probably because of the increased likelihood of quitting among those experiencing symptoms or diagnosed with illnesses.

A few studies examined reports of fatigue or tiredness. In a survey of New Zealand women who worked at home, smokers were 71 percent more likely than nonsmokers to report frequently feeling tired for no reason (Chetwynd and Rayner 1986). In a study of retired persons in the United States, after adjusting for age, current smokers were 60 percent more likely than lifetime nonsmokers to report becoming very tired easily (Rimer et al. 1990); former smokers were 25 percent more likely than lifetime nonsmokers to report getting very tired easily (Rimer et al. 1990).

Smokers tend to rate their general level of well-being lower than do nonsmokers whether well-being is measured directly (Dennerstein et al. 1994), assessed overall as quality of life (Sippel et al. 1999), or rated by degrees of general satisfaction with life (Blair et al. 1980) (Table 6.12). Similar findings have been observed when former smokers were compared with lifetime nonsmokers (Table 6.14) (Blair et al. 1980; Sippel et al. 1999). Conversely, compared with lifetime nonsmokers, current smokers tend to rate themselves as more dissatisfied with life (Table 6.12) (Kaprio and Koskenvuo 1988), but few differences in the prevalence rates of life dissatisfaction were observed between former smokers and nonsmokers (Table 6.14) (Kaprio and Koskenvuo 1988).

With respect to mental health and well-being, smokers tend to rate themselves slightly lower on measures of mental health or mental well-being (Wakefield et al. 1995; Wooden and Bush 1995; Sippel et al. 1999). In addition, smokers are more likely than nonsmokers to have psychological symptoms such as depressed mood and phobic anxiety (Matarazzo and Saslow 1960; Macnee 1991; Schoenborn and Horm 1993). In the South Wales study, not included in the summary tables, current smokers had a mean SF-36 mental health score that was slightly but not significantly lower than that of people who had never smoked (Lyons et al. 1994). Former smokers also tend to rate themselves less favorably than do nonsmokers (Table 6.14). The differences between former smokers and lifetime nonsmokers were small with respect to mental health and well-being (Wetzler and Ursano 1988; Wooden and Bush 1995; Sippel et al. 1999), but were more marked on measures of symptoms or morbidity (Table 6.14) (Lilienfeld 1959; Lindenthal et al. 1972; Macnee 1991). A strong dose-response trend was observed between smoking frequency and depressed moods in nationally representative U.S. data from the National Health Interview Survey (Schoenborn and Horm 1993). However, dose-response trends generally did not occur for mental health measures (Table 6.13) (Lindenthal et al. 1972; Wetzler and Ursano 1988; Stansfeld et al. 1993).

Studies of physical functioning, or functional status, among elderly populations also provide relevant evidence. Although they are not a focus of this review, such studies have provided prospective evidence that cigarette smoking is associated with accelerated declines in physical function (Pinsky et al. 1987; Guralnik and Kaplan 1989; Berkman et al. 1993; Strawbridge 1993). An analysis of data from the Honolulu Heart Study showed that smoking was inversely associated with freedom from clinical illnesses, physical impairment, and cognitive impairment (Reed et al. 1998).

The evidence provides a clear indication that smokers perceive their health as poorer than nonsmokers perceive theirs. Smokers report more symptoms (including mental health symptoms) and illness episodes, feel more tired, and have lower ratings for physical health status. Compared with nonsmokers, smokers even report lower overall levels of well-being for reasons that may at least partially reflect their diminished health status. The consistent indications of a poorer health status among smokers compared with nonsmokers across numerous health status dimensions provide direct evidence that smoking is associated with a diminished health status.

Evidence Synthesis

This section reviewed evidence on smoking and a diverse but interrelated set of measures of health status. Although the measures are nonspecific and likely to be affected by factors other than smoking, there is abundant and consistent evidence that smokers generally have a poorer health status than nonsmokers. This section reviewed findings on self-reported health statuses, absenteeism, and medical services utilization rates, as well as complications of surgical care. For each of these outcomes, the weight of the evidence indicates an adverse effect from smoking. There are many studies with differing designs and a variety of populations. The strength of the association with smoking is variable across the outcome measures and across study populations, probably reflecting the nonspecificity of these measures and the differing mixes of potential confounding and modifying factors across studies. In general, there is evidence for an increasing severity of outcome measures with an increasing number of cigarettes smoked, and current smokers tend to have worse outcomes than former smokers. Studies have addressed potential confounding factors to a limited extent, depending on the availability of data on relevant factors. Given the diversity of populations, study designs, and consistency of findings, confounding alone does not seem to be a satisfactory explanation for the overall pattern of findings. A single, unifying biologic basis for the association of smoking with the outcome measures cannot be postulated, but there are many well-supported direct and indirect mechanisms that may link smoking to the adverse effects documented in this section.

Conclusions

1. The evidence is sufficient to infer a causal relationship between smoking and diminished health status that may manifest as increased absenteeism from work and increased use of medical care services.

2. The evidence is sufficient to infer a causal relationship between smoking and increased risks for adverse surgical outcomes related to wound healing and respiratory complications.

Implications

Although preventing the specific diseases caused by smoking has been a public health priority for a long time, cigarette smoking also causes a substantial and costly burden of nonspecific morbidity. Smokers have a poorer health status, lose more time from work, and use medical care services at a higher rate than their nonsmoking peers. These adverse effects occur among younger smokers even before the burden of smoking-induced diseases becomes apparent at middle age and older.

Loss of Bone Mass and the Risk of Fractures

In the United States, of the estimated 850,000 fractures per year in persons 65 years of age and older, nearly 300,000 are hip fractures (Apple and Hayes 1994; Centers for Disease Control and Prevention [CDC] 1996; Ray et al. 1997). Approximately 33 percent of women and 17 percent of men experience a hip fracture if they live to be 90 years old (Mazess 1982; Melton and Riggs 1987). Mortality in persons with a hip fracture is 12 to 20 percent higher than in persons without a hip fracture of similar age, race, and gender (Miller 1978; Jensen and Tondevold 1979; Weiss et al. 1983; Jensen 1984; Kenzora et al. 1984; Kreutzfeldt et al. 1984). The estimated annual costs for medical and nursing services related to hip fractures range from $7 billion to $10 billion (Ray et al. 1997). From July 1991 through June 1992, costs to Medicare for 10 types of fractures were estimated at $4.2 billion (Baron et al. 1996). Moreover, continued growth of the elderly population can be expected to dramatically increase the number of hip fractures, because hip fracture incidence rates increase exponentially with age (Melton and Riggs 1987; Melton et al. 1987). If these demographic and incidence trends continue, the number of hip fractures may well double or triple by the middle of the century (Kelsey and Hoffman 1987). With their frequency, adverse quality of life impacts, and economic costs, hip fractures are an urgent and major public health problem.

Bone mineral density (BMD) is one of the strongest indicators of the risk for a fracture. Several cohort studies have confirmed that even a single low BMD measurement is associated with the risk of a later fracture (Gärdsell et al. 1989; Hui et al. 1989; Cummings et al. 1993). For each standard deviation decrease in BMD, the estimated relative risk (RR) of fractures ranged from 1.5 to 2.6, depending on the site that was measured (Marshall et al. 1996). Therefore, discussions of the possible adverse effects from smoking on bone health should consider both BMD and fractures as outcome measures. An estimated 60 to 80 percent of the bone density variation is explained by genetic factors (Eisman 1999), leaving 20 to 40 percent of the variation attributable to nongenetic factors. Smoking is an important modifiable risk factor in both women and men.

Conclusions of Previous Surgeon General’s Reports

Harmful effects of smoking on the skeleton have been recognized for several decades but the data were not sufficient to conclude that smoking adversely affects bone mass (USDHHS 1990); however, the most recent Surgeon General’s report on women and smoking (USDHHS 2001) identified smoking as adversely affecting bone health and increasing the risks for fractures. The report concluded that smoking adversely affects bone density and increases the risks for hip fractures in postmenopausal women. Specifically, the conclusions were that (1) postmenopausal women who currently smoke have lower bone density than women who do not smoke; (2) women who currently smoke have an increased risk for hip fracture compared with women who do not smoke; and (3) the relationship among women between smoking and the risk for bone fracture at sites other than the hip is not clear (USDHHS 2001). However, because male osteoporosis also has been recognized as a considerable disease burden, the role of smoking in male bone health also deserves consideration.

Biologic Basis

Smoking has the potential for direct and indirect effects on skeletal health and the risk of fractures. Direct toxic effects of smoking on bone cells may be related to the physiologic effects of nicotine (Fang et al. 1991; Riebel et al. 1995) or possibly cadmium in tobacco smoke (Bhattacharyya et al. 1988). Indirect effects of smoking on bone cells may result from decreased intestinal calcium absorption (Krall and Dawson-Hughes 1999), reduced intake and lower levels of vitamin D (Brot et al. 1999), or alterations in the metabolism of adrenal cortical and gonadal hormones (Michnovicz et al. 1986; Khaw et al. 1988; Baron et al. 1995). These direct and indirect effects may account for the generally observed decrease in markers of bone formation such as osteocalcin in smokers compared with nonsmokers (Brot et al. 1999; Bjarnason and Christiansen 2000). Smoking might also indirectly influence bone density through reduction in body weight, since body weight tends to be lower for smokers than for nonsmokers. This weight difference may itself lead to lower bone density and an increased risk for a fracture (Kiel et al. 1987; Cummings et al. 1995). Smokers also tend to have an earlier menopause than nonsmokers, thus extending the postmenopausal period of accelerated bone mineral loss (USDHHS 2001). Finally, smokers tend to be less physically active than nonsmokers and activity level is associated with bone density and hence risk for a fracture (Gregg et al. 1998).

In several analyses involving women, the lower weight of smokers compared with nonsmokers explains part of the increased risk for low BMD associated with smoking (Bauer et al. 1993). However, there are differences in BMD and in fracture rates between smokers and nonsmokers even after adjusting for weight differences, suggesting that the weight difference alone does not explain the effects of smoking (Kiel et al. 1992, 1996; Bjarnason and Christiansen 2000). The lower weight in smokers may increase the risk of fractures, such as hip fractures, through several mechanisms: reduced soft tissue mass overlaying the trochanter, resulting in less energy absorption from a fall on the hip; reduced weight loads on the skeleton; or reduced conversion of adrenal steroids into sex steroids in adipose tissue. The antiestrogenic effect of smoking also may contribute to osteoporosis in women (Jensen et al. 1985; Jensen and Christiansen 1988), and may reduce the benefits of hormonal replacement therapy (Komulainen et al. 2000). In a Finnish trial of osteoporosis prevention, smoking was associated with a nonresponse to hormonal therapy, as assessed by changes in BMD (Komulainen et al. 2000). Less consistent evidence for a blunted response to estrogen by smoking was reported from a Danish trial (Bjarnason and Christiansen 2000). Interestingly, although estrogen appears to be a critical hormone for male skeletal health (Slemenda et al. 1997; Khosla et al. 1998), smoking does not appear to modify the association between estradiol levels and bone density in men (Amin et al. 1999). Finally, smoking may increase the risk of fractures through reductions in physical performance capacity, thereby increasing the risk for falls (Nelson et al. 1994).

Bone Density in Young Men and Women

Epidemiologic Evidence

Increasingly refined measures of BMD have become available so that current studies use direct BMD measurements. Before such direct measurements were possible BMD was assessed using radiographs, with measurements typically focused on the widths of the cortical bones in sites such as the metacarpals. Direct quantitative assessments of the amount of mineral in various skeletal sites have now become possible with the advent of single and dual photon absorptiometry, followed by refinements such as single and dual x-ray absorptiometry, quantitative computed tomography, and quantitative ultrasonography. These techniques have all been used to generate the data summarized here.

In adults at any particular age bone mass is dependent on the peak mass achieved up to that age, and subsequent losses from the peak are attributable to aging and other factors. The pace of skeletal growth is rapid during infancy, slower during childhood, accelerated during puberty, and by 20 to 30 years of age the peak skeletal mass is attained (Kroger et al. 1992; Lu et al. 1996). Gains in BMD continue into the third decade after bone growth has ceased (Recker et al. 1992). After menopause, bone loss rates accelerate compared with premenopausal rates, and these rates are sustained or increase even more with aging (Ensrud et al. 1995). Age-related losses also occur in men (Jones et al. 1994). In the context of these age-related patterns, the role of smoking in the attainment of peak bone mass is reviewed along with studies of bone density and menopausal status. A literature search was conducted using the National Library of Medicine’s PubMed system; the key words used were “bone mineral density,” “bone density,” “fracture,” “smoking,” and “cigarettes.” In addition, all references from a key meta-analysis (Law and Hackshaw 1997) were also retrieved. Studies focusing on men mainly involve older age groups. The evidence on smoking and BMD comes primarily from cross-sectional and cohort studies. The cross-sectional studies assess the cumulative consequences of smoking on BMD growth and/or decline. Cohort studies can assess changes in BMD over time. Findings of the different types of studies are presented in Tables 6.15–6.17.

Table 6.15

Cross-sectional studies on the association between smoking status and bone density in women. 

Table 6.16

Studies on the association between smoking status and bone density in men and women published since the 1997 meta-analysis by Law and colleagues. 

Table 6.17

Cohort studies on the association between smoking status and the risk of bone loss in men and women. 

Peak Bone Mass

Because BMD increases rapidly during adolescence, initiating smoking around the time of puberty might reduce peak BMD. However, the effects of smoking on the attained level of peak bone mass are uncertain because there are limited data on the skeletal effects of smoking during adolescence. Furthermore, it is possible that relatively short exposures in this age group would have little effect on bone density measurements. One prospective cohort study of children and adolescents (aged 9 to 18 years) in Finland repeatedly ascertained lifestyle factors and followed participants for 11 years, at which time they underwent bone density testing (Välimäki et al. 1994). In men, but not in women, smokers had lower BMD measurements of the hip and spine than did nonsmokers after adjusting for covariates. A cross-sectional study of 15-year-old Swedish adolescents did not find an association between smoking and total body bone mineral content (Lötborn et al. 1999). Findings were similar in a cross-sectional study of 500 children aged 4 to 20 years in the Netherlands, but only 32 were smokers (Boot et al. 1997).

Data are available from studies of premenopausal women, starting from the ages at which peak BMD is reached. A meta-analysis of cigarette smoking, BMD, and the risk for hip fractures (Law and Hackshaw 1997) identified 10 cross-sectional studies of premenopausal women (Johnell and Nilsson 1984; McCulloch et al. 1990; Mazess and Barden 1991; Daniel et al. 1992; Fehily et al. 1992; Sowers et al. 1992; Hopper and Seeman 1994; Ortego-Centeno et al. 1994; Välimäki et al. 1994; Law et al. 1997). Additional study populations included menopausal and postmenopausal women (Table 6.15). As shown in Table 6.15, the mean ages of women in the study samples ranged from 22 to 76 years. Because absolute bone density units varied among studies according to the bone site assessed and the measurement technique used, the difference between the average BMD of current smokers and nonsmokers in each of the studies was recorded as a proportion of one between-person standard deviation. In combining the studies, each bone density difference was weighted by the inverse of its variance and was age-adjusted only.

Bone densities were reported for current smokers compared with never smokers in most studies, but were reported for current compared with former and lifetime never smokers combined in a few studies. There was no evidence of a significant difference in BMD between smokers and nonsmokers in the pre-menopausal women (Figure 6.2). Two additional studies of premenopausal and postmenopausal women performed since the 1997 meta-analysis also show no significant differences in BMD between smokers and nonsmokers (Table 6.16) (Takada et al. 1997; Gregg et al. 1999); however, a study of premenopausal women from Australia did find a significantly lower BMD in female current smokers that was not found in the subgroup of female smokers who participated in sports (Jones and Scott 1999). Cross-sectional data from the Danish Osteoporosis Prevention Study showed lower BMD in current smokers compared with lifetime non-smokers in perimenopausal women (Hermann et al. 2000). It is appropriate to consider these results unadjusted for other covariates in that adjusting for one of the most important risk factors for bone density— weight—actually may mask an association. Smoking-induced weight loss may represent an intervening variable in the causal chain between smoking and bone density reduction.

Figure 6.2

Differences (95% confidence intervals), as a proportion of 1 standard deviation (SD), in bone mineral density between female smokers and nonsmokers according to age and menopausal status. Note: Fitted regression lines are shown. The 11 open circles refer (more…)

One study from Spain assessed smoking and BMD in healthy young males (Ortego-Centeno et al. 1997). In this study, male volunteers aged 20 through 45 years were measured for BMD in the lumbar spine and proximal femur; blood biochemical markers were also assessed. BMD was significantly lower for smokers of 20 or more cigarettes per day compared with nonsmokers. In multiple regression analyses considering all smokers, smoking was not significantly associated with measures of BMD. Interpretations of these findings are limited by the cross-sectional data and the small sample size.

Smoking Cessation and Bone Mineral Density Loss

Two prospective cohort studies assessed smoking cessation and BMD in men and women (Hollenbach et al. 1993; Kiel et al. 1996). In a study in Rancho Bernardo, California, Hollenbach and colleagues (1993) found that smoking cessation later in life was beneficial for men and women in halting BMD loss at hip sites (intertrochanter, total hip, femoral neck, and trochanter) where BMD is reduced in smokers. In men, smoking cessation was followed by a reduction in the rate of loss of the spinal BMD, and women experienced a significant decrease in the rate of BMD loss at the midradius after quitting. In the Framingham study, current or former smoking (past 10 years) was not associated with a lower BMD loss at any skeletal site among women who had not taken estrogen but it was in women who had (Kiel et al. 1996). Former male smokers who had quit for less than 10 years had a lower BMD than men who had quit for 10 or more years, independent of weight, alcohol consumption, or caffeine use.

Evidence Synthesis

Smoking, even at a young age, might increase risk for osteoporosis later in life if it reduces the peak bone mass attained, thereby compromising the peak from which decline begins. Only a few studies address smoking during adolescence, and the findings in women during the premenopausal years are conflicting, are not based on large studies, and do not provide strong evidence for an effect of smoking on BMD before menopause. For males, data are scant for this age range. Although an effect of smoking on BMD is plausible, the available evidence from observational studies is limited and inconsistent.

Conclusion

1. The evidence is inadequate to infer the presence or absence of a causal relationship between smoking and reduced bone density before menopause in women and in younger men.

Implications

The failure to demonstrate a causal relationship between smoking and bone density in young women does not detract from the basis for concern about smoking and osteoporosis in women. For women, smoking patterns established in younger years are likely to persist past menopause, and there is substantial evidence linking smoking to low bone density during menopause (see below). Future research should quantify the combined and cumulative effects of premenopausal and postmenopausal smoking on bone density. More research is needed in young men regarding the relationship between smoking and bone density.

Bone Density in Middle and Later Years of Life

Epidemiologic Evidence

In contrast to the findings for younger persons, findings of bone density studies performed in populations well beyond the years of peak bone mass demonstrate substantial differences between smokers and nonsmokers. As illustrated in Figure 6.2, based on the meta-analysis by Law and Hackshaw (1997), bone density was lower in smokers than in nonsmokers for post-menopausal women, and the difference increased linearly with age. For every 10-year increase in age, the bone density of smokers fell below that of non-smokers by approximately 2 percent of the average bone density at the time of menopause, regardless of the skeletal site that was measured.

Since the publication of this meta-analysis, there have been additional studies of smoking and bone density in postmenopausal women and in men. Of four studies that did not demonstrate an association between smoking and bone density (Cheng et al. 1999; Varenna et al. 1999; Huuskonen et al. 2000; Kim et al. 2000), two had used quantitative ultrasound to measure bone status. Seven other studies did demonstrate statistically significant associations between smoking and BMD (Table 6.16) (Brot et al. 1997; Takada et al. 1997; Vogel et al. 1997; Grainge et al. 1998; Smeets-Goevaers et al. 1998; Hagiwara and Tsumura 1999; Hermann et al. 2000).

Data from cohort studies of older men and women also implicate smoking as a significant risk factor for bone loss (Table 6.17). Of the six studies that reported smoking data (three involving women and men, two involving women only, and one involving men only) (Sowers et al. 1992; Jones et al. 1994; Vogel et al. 1997; Burger et al. 1998; Guthrie et al. 1998; Hannan et al. 2000), three documented significantly more bone loss in female smokers than in female and male nonsmokers (Sowers et al. 1992; Burger et al. 1998; Guthrie et al. 1998), and three reported higher rates of loss among male smokers than among male nonsmokers (Vogel et al. 1997; Burger et al. 1998; Hannan et al. 2000). Interpretations of several of the studies are constrained by relatively small sample sizes and limited durations of follow-up.

Evidence Synthesis

Extensive and consistent data are available on BMD and smoking for perimenopausal and postmenopausal women and for older men. Data from cohort studies, which track changes in BMD over time, as well as from cross-sectional studies provide generally consistent evidence of increased rates of loss in postmenopausal women who smoke compared with nonsmokers. Smoking cessation appears to benefit BMD since limited data indicate higher rates of BMD loss for heavier smokers. Data are more limited for men. The 2001 Surgeon General’s report (USDHHS 2001) found the evidence to be consistent for women and concluded that “Postmenopausal women who currently smoke have lower bone density than do women who do not smoke” (p. 321). There are a number of mechanisms that may underlie this finding.

Conclusions

1. In postmenopausal women, the evidence is sufficient to infer a causal relationship between smoking and low bone density.

2. In older men, the evidence is suggestive but not sufficient to infer a causal relationship between smoking and low bone density.

Implications

Smoking has an adverse effect on bone density in middle and later years of life; for every 10-year increase in age, the bone density of female smokers falls below that of nonsmokers by about a 0.14 standard deviation, or 2 percent of the average bone density at the time of menopause in women. Because a 1.0 standard deviation decrease in bone density doubles the risk of fracture, and because fracture incidence increases with age (Melton and Riggs 1987; Melton et al. 1987), the proportion of all fractures attributable to smoking would be expected to increase for smokers who continue smoking into older ages. Attempts to decrease smoking as early in life as possible are likely to reduce fractures that would be caused by smoking in old age.

Because bone loss is relatively small over short periods of time, studies with longer durations of follow-up and minimal avoidable losses of participants at follow-up could add important information to the understanding of how smoking contributes to bone loss. Additional information is likely to come from studies of biochemical markers of bone turnover, which might further the understanding as to mechanisms whereby smoking accelerates bone loss.

Fractures

Epidemiologic Evidence

Hip fractures, the most frequently studied fractures in relation to smoking, account for a significant proportion of the morbidity and mortality attributed to osteoporosis. The meta-analysis by Law and colleagues (1997) reviewed 19 cohort and case-control studies of the risk of hip fractures in postmenopausal women according to whether they had COPDs. The studies differed with regard to the ages of the participants, duration of follow-up, and whether former smokers were included in the smoking or nonsmoking groups. Table 6.18 shows the characteristics of each of the 19 studies, demonstrating the range of ages at the time of the fracture. For the cohort studies, the duration of follow-up ranged from three years (Forsén et al. 1994) to 26 years (Kiel et al. 1992). Figure 6.3 shows the risk of hip fractures in smokers relative to non-smokers according to age; the risks for smokers increased with increasing age. Major conclusions of the meta-analysis include (1) smoking has no material effect on bone density in premenopausal women; (2) postmenopausal bone loss is greater in smokers—an additional 0.2 percent of bone mass each year; (3) in comparisons of women who are current smokers with women who are nonsmokers, the risk of hip fracture is estimated to be 17 percent greater at 60 years of age, 41 percent greater at 70 years, 71 percent greater at 80 years, and 108 percent greater at 90 years; and (4) the estimated cumulative risk of hip fracture to 85 years of age in women is 19 percent in smokers and 12 percent in nonsmokers; to 90 years it is 37 percent and 22 percent, respectively. The data for men were much more limited but suggested similar consequences.

Table 6.18

Studies on the association between smoking and the risk of hip fractures in men and women used in the 1997 meta-analysis by Law and Hackshaw. 

Figure 6.3

Relative risk (95% confidence intervals) of hip fracture in smokers compared with nonsmokers in postmenopausal women according to age. Note: Each cohort study (8 solid circles) and case-control study (11 open circles) is in the same order as in Table (more…)

Since the publication of the meta-analysis by Law and colleagues (1997), some (Forsén et al. 1998; Burger et al. 1999; Kanis et al. 1999; Melhus et al. 1999; Baron et al. 2001) but not all subsequent studies of hip fracture (Fujiwara et al. 1997; Clark et al. 1998; Mussolino et al. 1998) have continued to show an association between smoking and an increased risk of hip fracture (Table 6.19). These studies have used various designs and have been carried out in diverse populations.

Table 6.19

Studies on the association between smoking and the risk of hip fractures in men and women reported since the 1997 meta-analysis by Law and Hackshaw. 

Data on the association between smoking and fractures at other sites are more limited (Table 6.20). Studies from the 1980s and early 1990s that examined fractures other than those of the hip rarely found an association with smoking, although more recent studies have demonstrated positive associations between smoking and vertebral fractures (Scane et al. 1999; Lau et al. 2000), ankle fractures (Honkanen et al. 1998), and the general categories of nonhip fractures (Jacqmin-Gadda et al. 1998) and of all fractures (Huopio et al. 2000).

Table 6.20

Studies on the association between smoking and the risk of fractures at sites other than the hip in men and women. 

Smoking Cessation and Hip Fractures

The association between smoking cessation and the risk of hip fractures was examined in several studies, including three prospective cohort studies with follow-up periods of 5 to 12 years (Forsén et al. 1998; Cornuz et al. 1999; Høidrup et al. 2000) and two case-control studies (La Vecchia et al. 1991; Cumming and Klineberg 1994). In men, successful smoking cessation of at least five years decreased the risk of hip fracture compared with continuing smokers (Høidrup et al. 2000), although other investigations found that this risk remained elevated for men and women smokers compared with lifetime nonsmokers (Cumming and Klineberg 1994; Forsén et al. 1998). Two studies also found no decrease in the risk for hip fractures in women after five years of smoking cessation (La Vecchia et al. 1991; Cornuz et al. 1999), and another found that no benefit from quitting for women, including premenopausal women, was observed until 10 years after cessation (adjusted RR = 0.7 [95 percent confidence interval (CI), 0.5–0.9] compared with current smokers) (Cornuz et al. 1999).

Evidence Synthesis

The evidence on smoking and fracture has been reviewed extensively in previous reports of the Surgeon General. The 1990 report considered evidence from eight case-control studies, noting that most showed an association with risk for fracture of the hip or vertebra. Five cohort studies, however, did not show a clear increase in risk and the report found the evidence to be inconclusive. Far more extensive data were available for the 2001 report, including substantially more studies of hip fracture in women. The case-control studies reviewed all indicated excess risk for hip fracture in smokers, with the RR ranging from 1.1 to 2.0. Six reports of cohort studies published subsequent to the 1990 report were also cited, all showing an increased risk for hip fracture in current smokers. The 2001 report (USDHHS 2001) concluded that “women who currently smoke have an increased risk for hip fracture compared with women who do not smoke” (p. 321).

This report extends the review of the 2001 report with additional studies and covers the evidence on men as well. The evidence consistently indicates an increased risk for women and men who smoke. Findings of some studies show a dose-response relationship between risk for hip fracture and the amount smoked. The RR tends to rise with age as would be expected, and the effect of smoking reflects sustained, additional bone loss beyond that associated with aging. The documented effects of smoking on BMD is consistent with the observational evidence on hip fracture.

For fracture sites other than the hip, the evidence has been less consistent. The 2001 Surgeon General’s report found the evidence to be unclear. This report evaluated a number of studies for other sites, also finding the evidence to be mixed and limited in scope for any particular site.

Conclusions

1. The evidence is sufficient to infer a causal relationship between smoking and hip fractures.

2. The evidence is inadequate to infer the presence or absence of a causal relationship between smoking and fractures at sites other than the hip.

Implications

The RR of hip fractures in smokers increases with age, and hip fracture incidence increases with age, implying that the proportion of hip fractures attributable to smoking increases with age. Smoking is one of the major causes of fracture in older persons that can be prevented. Public health interventions aimed at helping smokers quit are likely to substantially reduce the number of hip fractures. Although hip fractures carry the greatest costs and risks of mortality and morbidity, other fractures also contribute to these outcomes. Further research is necessary to quantify the risks of these other fractures in smokers.

Go to:

Dental Diseases

Diseases of the teeth and their supporting structures are a major public health issue with a significant impact on personal well-being. More than $60 billion were spent on oral health care in the United States in 2000, and each year acute oral conditions result in an estimated 1.6 million missed school days and 2.4 million lost workdays. Although there have been tremendous improvements in the oral health of the U.S. public during the past several decades, oral diseases and conditions remain highly prevalent. For example, recent national data indicate that 66 percent of persons aged 12 through 17 years and 94 percent of those aged 18 years and older have experienced dental caries in their permanent teeth (USDHHS 2000).

As the oral cavity is the first part of the human anatomy to be exposed to mainstream smoke in active smokers, researchers have long hypothesized that smoking could have a deleterious effect on the teeth and their supporting structures. However, research on this association was hampered for decades by (1) lack of consensus on case definitions for some diseases; (2) difficulty in measuring oral conditions and consequent use of indices of questionable validity; (3) some incorrect assumptions about disease etiology, pathogenesis, distribution, and natural history; and (4) limited capacity for epidemiologic investigations within the dental research community. As a result, until recently the literature was sparse and findings were not definitive.

Conclusions of Previous Surgeon General’s Reports

The previous Surgeon General’s reports on smoking and health did not include dental or periodontal effects of smoking, although oral cancer and related premalignant lesions have been addressed. During the past 15 years, however, there has been a substantial amount of research on smoking and oral health, and this topic was addressed in Oral Health in America: A Report of the Surgeon General (USDHHS 2000). This section reviews the epidemiologic evidence for smoking as a causal factor for the most common forms of non-malignant oral disease; cancers of the oral cavity are covered in Chapter 2.

Periodontitis

The periodontium includes those hard and soft tissue structures that support the teeth: the gingiva, the cementum covering the root surfaces of the teeth, the periodontal ligament that attaches the tooth root surfaces to the adjacent alveolar bone supporting each tooth, and the alveolar bone. The gingiva covers the other periodontal structures and comprises attached and free gingiva. The attached gingiva extends from the bottom of the gingival sulcus to the mucogingival junction, where it is contiguous with the mucous membrane of the lip, cheek, and floor of the mouth. The free gingiva extends from the base of the gingival sulcus to the gingival margin.

In a healthy state, the gingival margin is approximately 0.5 to 2.5 mm coronal to the cemento-enamel junction (CEJ) (where the enamel on the crown of the tooth meets the root). The sulcus is 1 to 3 mm in depth and does not bleed when probed. The base of the sulcus is formed by the junctional epithelium, which joins the gingival connective tissue to the tooth surface. Healthy gingiva is usually pink in color, is well adapted to the teeth, has a stippled surface texture, and is tightly bound to the underlying alveolar bone and the roots of the teeth.

Based on the most recent classification system developed by the American Academy of Periodontology, there are at least eight categories of periodontal diseases and conditions (Armitage 1999). Of those, the two most common are gingivitis and chronic periodontitis. Gingivitis is defined as an inflammation of the gingiva in which the junctional epithelium remains on or near the enamel covering the crown of the tooth. It is characterized clinically by redness, gingival bleeding, edema or enlargement, and occasional gingival sensitivity and tenderness (Genco 1990a). Chronic periodontitis (previously called adult periodontitis) is an inflammation of the gingiva and the adjacent attachment apparatus that is characterized by loss of clinical attachment because of destruction of the periodontal ligament and loss of the adjacent supporting bone (Flemmig 1999). Clinical features of chronic periodontitis may include edema, erythema, gingival bleeding upon probing, periodontal pocketing, or suppuration.

The most common forms of both gingivitis and periodontitis involve bacterial infection. Severe forms of periodontitis often are associated with infection by specific bacteria that colonize the subgingival area (Genco 2000). Destruction of soft tissue and alveolar bone is thought to involve toxins and proteases produced by the bacteria as well as hyperresponsiveness and reactivity of various components of the immune system (e.g., the production of cytokines and prostaglandins). Smoking may play a role in the pathogenesis of periodontal diseases by altering immune function and tissue repair.

The understanding of the distribution and natural history of periodontitis has evolved over the past several decades. Previously, it was thought that virtually all persons were susceptible to severe disease if oral hygiene was inadequate. The disease was considered to progress in a linear fashion throughout life from gingivitis to periodontitis to bone loss to tooth loss, generally attacking the entire dentition and was nearly universal among adults (World Health Organization 1961). This concept was driven, in part, by epidemiologic indices that incorporated signs of both gingivitis and periodontitis, analytic methods that aggregated and averaged measurements within persons and populations, and assumptions about disease progression on the part of the early oral epidemiologists. In the current model of periodontal diseases, a small proportion of persons in most populations are considered to have severe periodontitis; periodontitis is usually preceded by gingivitis but few sites with gingivitis later develop periodontitis; periodontal tissues can undergo some degree of self-repair; and generalized forms of periodontitis are uncommon (American Academy of Periodontology 1996; Burt and Eklund 1999).

Based on current concepts of periodontitis, clinical or epidemiologic assessment of the disease involves detailed measurements of various signs of soft tissue or bone destruction at two to six sites per tooth either on all teeth or on selected teeth. Among the most common measurements is probing pocket depth (PPD), which is measured by inserting a calibrated probe into the gingival sulcus and recording the distance in millimeters from the gingival margin to the base of the gingival sulcus (if healthy) or pocket (if diseased). Because the pathogenesis of periodontitis involves destruction of the junctional epithelium at the base of the sulcus, a PPD greater than 4 mm may indicate disease (Genco 1990b). Another common parameter is the clinical attachment level (CAL), which is measured as distance in millimeters from the CEJ to the base of the gingival sulcus or pocket. It is a direct measure of the position of the periodontal epithelial attachment of a tooth relative to its ideal position at the CEJ. Many cross-sectional studies have used the terminology “loss of periodontal attachment” (LPA) to describe this same parameter, although more recent studies tend to reserve the use of the term LPA for longitudinal assessments of change in the CAL between two points in time. The longitudinal change in CAL is sometimes called relative attachment loss, particularly when computer-linked electronic periodontal probes are used to record the measurements from a fixed reference point such as a cusp tip. Examples of all of these parameters and terms are found in the epidemiologic literature on the association between smoking and periodontal destruction. Because periodontal destruction may occur without deep pocket formation, PPD alone will underestimate disease and may not be sufficient as the prime indicator of disease (Goodson 1990). Intraoral radiographs have been used to assess alveolar bone loss from periodontitis, but this approach can have low sensitivity and may underestimate true bone loss (Goodson 1990; Eickholz and Hausmann 2000; Pepelassi et al. 2000). In addition, radiography often is not logistically feasible or acceptable to examinees during large-scale field epidemiologic studies. At this time, change in the CAL is considered the prime indicator of periodontal destruction.

Biologic Basis

Microbiology

It is possible that cigarette smoking affects periodontal health by altering the quantity or composition of bacterial dental plaque. Although some studies found that smokers had more visible bacterial plaque than nonsmokers (Sheiham 1971; Bastiaan and Waite 1978; Lavstedt et al. 1982; Preber and Bergström 1985), many other studies reported no significant differences in mean plaque levels or rates of plaque accumulation (Alexander 1970; Swenson 1979; Bergström 1981, 1990; Feldman et al. 1983; Macgregor et al. 1985; Bergström and Eliasson 1987a,b; Lie et al. 1998). Cross-sectional differences in plaque levels between smokers and nonsmokers may be due to differences in oral hygiene practices rather than to smoking per se (Preber and Kant 1973; Andrews et al. 1998). However, the presence of specific bacterial species in periodontal plaque may be more important than the quantity of visible plaque and debris on the teeth in the pathogenesis of severe periodontitis (Genco 1996). Some evidence indicates that smokers may be more likely than nonsmokers to harbor specific periodontal pathogens. A study of adults exhibiting a wide range of periodontal conditions (Zambon et al. 1996) found that subgingival infection with Bacteroides forsythus was more common in current smokers even after adjusting for disease severity, with a dose-response relationship between the amount of smoking and infection. Current smokers were also more likely than former or lifetime nonsmokers to have subgingival infection with Actinobacillus actinomycetemcomitans. Consistent with those findings, a study of dental clinic patients found that plaque samples from smokers were 11 times more likely than samples from nonsmokers to test positive for one of three periodontal pathogens (Kazor et al. 1999). In a study of young adults with early-onset periodontitis (Kamma et al. 1999), 11 postulated periodontal pathogens were detected more frequently and in greater numbers in the subgingival plaque from smokers than from nonsmokers. Smoking may increase the likelihood of infection with periodontal pathogenic microorganisms even among persons with no clinical signs of disease. In a study of young adults who did not have periodontitis (Shiloah et al. 2000), smokers were 18 times more likely than nonsmokers to have at least one of eight periodontal pathogens in their sub-gingival plaque. Several studies, however, reported no differences in the plaque bacteria between smokers and nonsmokers (Preber et al. 1992; Stoltenberg et al. 1993). Additional evidence suggests that smoking may act synergistically to potentiate the effects of toxins produced by periodontal pathogenic bacteria (Sayers et al. 1999).

Immune Function

There is substantial evidence that smoking affects both localized and systemic components of the immune system, although the links between these effects and periodontal disease remain to be established. Smoking increases the number but impairs the functions of polymorphonuclear leukocytes (PMNs, or neutrophils), peripheral blood cells that represent the first line of defense against microorganisms (Noble and Penny 1975; Barbour et al. 1997). Either an impairment of the PMN’s ability to neutralize periodontal infections or an overstimulation of potentially tissue-destructive processes can lead to periodontal destruction (American Academy of Periodontology 1999). For example, smoking can impair PMN chemotaxis, phagocytosis, and oxidative burst (Eichel and Shahrik 1969; Kenney et al. 1977; Ryder et al. 1998). Impaired phagocytosis has been implicated in refractory periodontitis (MacFarlane et al. 1992). Smoking also appears to compromise the function of macrophages, which play a vital role in both humoral and cell-mediated immunity, and of B lymphocytes, the major cell type involved in the humoral immune system. Exposure to cigarette smoke also appears to have an immunosuppressive effect on T lymphocytes, which may reduce antibody response to periodontal bacteria (Barbour et al. 1997). Smokers may have a decreased production of antibodies specific to periodontal pathogens, especially IgG2 (Quinn et al. 1998). Recent evidence suggests that levels of cytokines in gingival crevicular fluid, which are secreted by mononuclear cells and are associated with collagen destruction and bone resorption, may be increased in smokers (Boström et al. 1998a,b). Furthermore, there may be a synergistic interaction between smoking and the genotype for a specific cytokine, IL-1, in the development of severe periodontitis (Kornman and di Giovine 1998).

Gingival Blood Flow and Soft Tissue Effects

It has long been hypothesized that the peripheral vasoconstrictive effect of tobacco smoke and nicotine reduces gingival blood flow and thereby impairs the delivery of oxygen and nutrients to gingival tissue. There is some evidence of reduced blood flow in gingival tissues (Clarke et al. 1981; Clarke and Shephard 1984) and reduced size and altered morphology of capillaries in oral mucosa and gingival tissues (Johnson et al. 1989) following exposure to tobacco smoke or nicotine. However, more recent evidence appears contradictory (Baab and Öberg 1987; Johnson et al. 1991). Smokers tend to exhibit less gingival bleeding than nonsmokers, even with control for bacterial plaque levels (Preber and Bergström 1985, 1986; Bergström and Preber 1986; Bergström 1990; Danielsen et al. 1990; Newbrun 1996). However, this reduced gingival bleeding may be related more to the suppression of an inflammatory response than to reduced gingival blood flow.

Nicotine can be stored in and released from periodontal fibroblasts, possibly affecting their morphology and ability to attach to root surfaces (Raulin et al. 1988; Hanes et al. 1991; James et al. 1999). In addition, nicotine may inhibit the growth of gingival fibroblasts and their production of collagen and fibronectin, components of the gingival extracellular matrix involved in the structure and attachment of gingiva (Tipton and Dabbous 1995). Thus, it is possible that smoking impairs the ability of periodontal tissues to repair damaged junctional epithelium. Smoking impairs wound healing and compromises the prognosis following surgical and nonsurgical periodontal therapy (Preber and Bergström 1990; Ah et al. 1994; Newman et al. 1994; Rosenberg and Cutler 1994; Preber et al. 1995; Tonetti et al. 1995; Grossi et al. 1996, 1997; Kaldahl et al. 1996; Kinane and Radvar 1997; Trombelli and Scabbia 1997; Boström et al. 1998b; Machtei et al. 1998; Renvert et al. 1998; Palmer et al. 1999; Papantonopoulos 1999; Söder et al. 1999). One study that employed statistical modeling of longitudinal changes in the CAL concluded that diminished capacity for repair, rather than direct tissue damage, probably was the major mechanism involved in smoking-associated periodontal destruction (Faddy et al. 2000).

Epidemiologic Evidence

Epidemiologic studies of smoking and periodontitis have employed a variety of case definitions for disease, using various combinations of PPD, CAL or LPA, and alveolar bone loss. Some studies used indices for “periodontal disease” that are no longer considered valid indicators for the prevalence of disease in populations (Burt and Eklund 1999). Other studies employed indices that originally were intended for use in population-based treatment planning and not for etiologic studies, such as the Community Periodontal Index of Treatment Needs (Ainamo et al. 1982). Some studies did not use a case definition for disease, but instead assessed mean levels of one or more clinical parameters among exposed and unexposed groups, or described the proportion of the study population that exceeded various measurement thresholds (e.g., ≥4 mm LPA). Some studies, primarily conducted before the 1970s, provided no case definition other than diagnosis by the examiner. Despite the numerous problems measuring the disease, published epidemiologic and clinical studies consistently show a moderate to strong degree of association between smoking and periodontitis.

To identify epidemiologic studies of smoking and periodontitis, the National Library of Medicine’s PubMed database was searched for English language publications from 1965–2000, using the following Medical Subject Headings (MeSH) key words: “smoking,” “tobacco,” “periodontal diseases,” and “periodontitis.” These terms also were searched as title words. The smoking and health database maintained by the Office on Smoking and Health, National Center for Chronic Disease Prevention and Health Promotion, CDC, was also searched using those terms as key words. Reference lists from published studies, review articles, and textbooks were examined to identify additional studies.

Tables 6.21 through 6.23 summarize the findings from 6 case-control studies, 52 cross-sectional studies, and 12 cohort studies conducted between 1959 and 2000. The case-control studies consistently found that persons with periodontitis were more likely than controls without periodontitis to be smokers, although not all studies separated current smokers from former smokers in their analyses. These studies generally controlled for potential confounders in either the selection of a control group or in their analyses. Cross-sectional studies that attempted to estimate parameters such as the odds ratio (OR) consistently reported moderate to strong degrees of association between smoking and periodontitis under a wide range of case definitions (Beck et al. 1990; Horning et al. 1992; Haber et al. 1993; Stoltenberg et al. 1993; Grossi et al. 1994, 1995; Sakki et al. 1995; Tomar et al. 1995; Ahlberg et al. 1996; Dolan et al. 1997a; Norderyd and Hugoson 1998; Shizukuishi et al. 1998; Wakai et al. 1999; Tomar and Asma 2000). Consistent with the findings from case-control and cross-sectional studies, cohort studies reported RR estimates for smoking and onset or progression of periodontitis of 1.4 to more than 10, using a wide range of outcome measures. Of the cross-sectional studies that examined the relationship separately for current smokers and former smokers, current smokers were more likely than former smokers to have periodontitis (Haber et al. 1993; Dolan et al. 1997a; Wakai et al. 1999; Tomar and Asma 2000). Two case-control studies (Haber and Kent 1992; Gelskey et al. 1998) and several cross-sectional studies (Grossi et al. 1994, 1995; Norderyd and Hugoson 1998; Wakai et al. 1999; Tomar and Asma 2000) reported a significant dose-response relationship between the number of cigarettes smoked per day and disease status. Two of these studies used cigarette-years2 or pack-years as the measure for exposure (Grossi et al. 1994, 1995), which combined quantity and duration of smoking to characterize the exposure. One study reported a significant dose-response relationship between the duration of smoking and disease risk (Tomar and Asma 2000). That study also found a significant inverse relationship between the number of years since quitting smoking and the odds of having periodontitis.

Table 6.21

Case-control studies on the association between smoking and periodontitis. 

Table 6.22

Cross-sectional studies on the association between smoking and periodontitis. 

Table 6.23

Cohort studies on the association between smoking and periodontitis. 

Nearly all other reviewed studies reported either mean measures of PPD or CAL/LPA or radiographically demonstrated alveolar bone loss by smoking status, or they reported the percentage of persons with some specified number or percentage of sites exceeding some threshold on one or more of these clinical parameters. With only one exception (Preber et al. 1980), all cross-sectional and cohort studies that measured differences in mean CAL/LPA or mean PPD found a worse periodontal status among smokers than among nonsmokers. That 1980 study (Preber et al. 1980), however, was conducted with young military recruits whose duration of smoking must have been relatively short because of their age.

Evidence Synthesis

The available epidemiologic literature is highly consistent in showing a moderate to strong association between cigarette smoking and periodontal destruction. The association is robust across a wide range of case definitions, populations, and study designs. There is also evidence of a dose-response relationship between smoking intensity and risk for periodontitis. Both number of cigarettes smoked and duration of smoking are positively associated with disease risk. The risk of periodontitis appears to decrease after smokers stop smoking, with a decreasing risk as the duration of successful cessation increases. Although only a few prospective cohort studies have been carried out, they consistently found that smokers were more likely than nonsmokers to experience the onset or progression of disease. The association cannot be explained by confounding.

The mechanisms involved in smoking-associated periodontal destruction are still not fully understood. However, available evidence supports several hypotheses. An immune mechanism is plausible because smoking affects many elements of the human immune system. The effects of smoking on local and systemic immune factors may make the smoker more susceptible to bacterial infection. In addition, substantial evidence indicates that smoking impairs the regeneration and repair of periodontal tissues. The evidence is inconsistent in suggesting that smoking quantitatively or qualitatively alters the microflora of subgingival plaque.

Conclusion

1. The evidence is sufficient to infer a causal relationship between smoking and periodontitis.

Implications

Smoking intervention should be a major component of prevention and treatment of periodontitis. A recent study (Tomar and Asma 2000) concluded that more than 50 percent of the cases of adult periodontitis in the United States are attributable to cigarette smoking. In light of this conclusion, and because more than one-half of U.S. adult smokers visit a dentist each year (Tomar et al. 1996), the dental care community has both the opportunity and the professional obligation to counsel patients who smoke to quit. The dental office may also provide an opportune setting for tobacco use prevention efforts among young people (Hovell et al. 1996). Unfortunately, a lack of awareness and inadequate skills may be barriers to further involvement by dentists and dental hygienists (Secker-Walker et al. 1994; Dolan et al. 1997b).

Further research is needed to achieve a greater understanding of the mechanisms involved in smoking-associated periodontitis. In addition, more behavioral research is needed to enhance the willingness and ability of dentists and dental hygienists to intervene in their patients’ use of tobacco and to counsel younger patients against tobacco use. Educational research should identify effective methods for training students of dentistry and dental hygiene, as well as licensed clinicians, to become competent at counseling their patients to stop using tobacco and assisting patients who want to quit (Tomar et al. 1996; Barker and Williams 1999; Cabana et al. 1999).

Dental Caries

Dental caries is an infectious, communicable, multifactorial disease in which bacterially produced acids dissolve the hard enamel surface of a tooth (Featherstone 1999). Unchecked, the bacteria may then penetrate the underlying dentin and progress into the soft pulp tissue, which is rich in blood and nerve tissue. Dental caries commonly results in loss of tooth structure and discomfort. Untreated dental caries commonly progresses to incapacitating pain and a bacterial infection that leads to pulpal necrosis, tooth extraction, and loss of dental function, and can progress to an acute systemic infection. The major etiologic factors for this disease are thought to be specific bacteria in dental plaque (particularly Streptococcus [S.] mutans and S. lactobacilli) on susceptible tooth surfaces and the availability of fermentable carbohydrates.

Most epidemiologic studies conducted during the past 60 years have used some variation of the decayed, missing (due to caries), or filled permanent teeth (DMFT) index (Klein et al. 1938) to measure the frequency of dental caries. Until the mid-1980s the proportion of the population with dental caries was rarely used to estimate disease prevalence in industrialized populations because the disease was nearly universal. The DMFT index is more a measure of disease severity than of disease prevalence; it is simply the sum of the number of permanent teeth (T) that are decayed (D), missing due to dental caries (M), or filled (F). This index, if applied to the number of coronal (i.e., enamel-covered) tooth surfaces (S), is designated the DMFS. The M component is often omitted in adult studies because of the inherent uncertainty as to why a tooth is missing. Thus, some studies report DFT or DFS scores. Other studies report the components of DMFT individually, such as DS, FS, and MS. Nearly all studies aggregate DMF data by reporting the population mean. The number of root surfaces affected by caries is almost always scored and reported separately from coronal caries, and usually is designated as RDFS or RDS (the M component is not reported for root-surface caries).

Biologic Basis

There are several hypothesized mechanisms that may underlie the association between smoking and dental caries. As discussed in the section on smoking and periodontitis, evidence is inconsistent in showing that smoking per se alters either the bacterial profile in the gingivi or the rate of formation of dental plaque (Alexander 1970; Swenson 1979; Bergström 1981, 1990; Feldman et al. 1983; Macgregor et al. 1985; Bergström and Eliasson 1987a,b; Lie et al. 1998). Differences in oral care behavior between smokers and nonsmokers provide an indirect explanation. Perhaps the most consistent explanation is that smokers tend to practice less frequent or less effective oral hygiene and plaque removal (Preber and Kant 1973; Macgregor and Rugg-Gunn 1986; Andrews et al. 1998).

Several studies concluded that smoking might lower the pH or reduce the buffering capacity of saliva (Heintze 1984; Parvinen 1984), impairing the function of saliva as a protective factor against enamel demineralization (Edgar and Higham 1996). In contrast, one review concluded that smoking increases salivary flow rate (Macgregor 1989), raising pH and increasing salivary calcium concentration (ten Cate 1996). These factors would tend to favor enamel remineralization, but benefit would come only if the flow rate increase were sustained. Another comprehensive review concluded that smoking has a minor effect on saliva flow rate and its chemical composition, at least in terms of factors thought to affect dental cariogenesis (Christen et al. 1991). In sum, an effect of smoking on salivary function does not appear to be a key mechanism in causing dental caries.

The association between smoking and root-surface caries suggested by several studies may be due, in part, to the periodontal effects of smoking. The loss of periodontal attachment and subsequent exposure of root surfaces are necessary conditions for root-surface caries to occur (Burt et al. 1986; Stamm et al. 1990). Persons who experience a loss of periodontal attachment attributable to smoking may also be at greater risk for subsequent root-surface caries.

Epidemiologic Evidence

To identify the epidemiologic studies on smoking and dental caries, the National Library of Medicine’s PubMed database was searched for English language publications from 1965–2000. The following MeSH key words were used: “smoking,” “tobacco,” “dental caries,” and “tooth demineralization.” These terms also were searched as title words. The smoking and health database maintained by CDC’s Office on Smoking and Health was also searched using the same terms as key words. Reference lists from published studies, review articles, and textbooks were sources for additional studies.

Table 6.24 summarizes 12 cross-sectional studies and 3 cohort studies published between 1952 and 1999. Most cross-sectional studies used some variation of the DMF index to measure caries prevalence; all but two (Hart et al. 1995; Tomar and Winn 1999) found that smokers experienced more coronal dental caries than nonsmokers, as measured by mean DS, DFS, DMFS, or DMFT. In general, differences between smokers and nonsmokers in mean DMFT or DMFS were small, even in studies in which the differences were reported to be “statistically significant.” The largest differences in numbers of carious lesions were reported in studies that used DMFS (Ludwick and Massler 1952; Ainamo 1971; Zitterbart et al. 1990; Axelsson et al. 1998). None of those studies, however, appeared to limit the “missing” component of DMFS to those tooth surfaces lost due to caries. Consequently, these studies may mix caries caused by smoking with the advanced periodontal destruction that can cause tooth loss in adults.

Table 6.24

Cross-sectional and cohort studies on the association between smoking and dental caries. 

Few of the studies on the association between smoking and dental caries controlled for potential confounding factors. Although the observed association between smoking and dental caries may reflect a causal relationship, it is also possible that it reflects factors common to both smoking and the risk of dental caries. For example, in industrialized nations both dental caries (USDHHS 2000) and cigarette smoking (Giovino et al. 1995) are more prevalent among groups with lower socioeconomic status (SES) than among higher SES groups. SES is a strong correlate of factors that affect dental caries status, such as diet, use of dental services, and oral hygiene practices (USDHHS 2000). None of the studies adjusted for SES or other potential confounding factors in examining the association between smoking and dental caries. Several literature reviews do suggest that the association between smoking and dental caries may reflect the tendency for smokers to practice less effective dental hygiene and plaque removal (Macgregor 1989; Christen et al. 1991; Kassirer 1994; Andrews et al. 1998).

Few studies adjusted for other notable correlates of both smoking and dental caries in their analyses. The DMF index is a cumulative, irreversible index. As persons experience decayed or filled permanent tooth surfaces or lose teeth over their lifetimes, their DMFT or DMFS scores will increase. Therefore, DMFT and DMFS can be associated strongly with age even if age per se is not a risk factor for incidence of dental caries. Few studies, however, adjusted for age in their analyses. Several studies provided age-specific mean caries scores (Ludwick and Massler 1952; Zitterbart et al. 1990; Axelsson et al. 1998) or age-specific significance testing of differences in means (Hirsch et al. 1991), which revealed an inconsistent association between smoking and caries within age groups. In the one study that used a nationally representative sample of U.S. adults and adjusted for age and race or ethnicity, DFT and DMS were actually slightly lower among male smokers than among those who had never used tobacco (Tomar and Winn 1999).

Two studies attempted to investigate a dose-response relationship between smoking and dental caries (Ludwick and Massler 1952; Ainamo 1971). Although smokers in the highest category of cigarettes smoked per day had experienced slightly higher DMFT, DMFS, or DS than those in the lowest dose categories, the relationship was not consistent. The first study presented age-specific comparisons of mean DMFT and DMFS by the number of cigarettes smoked per day, which showed no clear pattern within age strata. The second study did not present age-stratified or age-adjusted estimates, which potentially could present difficulties in interpreting the association between a disease index that is cumulative with age and an exposure that probably was increasing with age in the study population (aged 18 through 26 years).

Smoking may be associated more with root-surface caries than with coronal caries. Two cohort studies (Ravald et al. 1993; Locker 1996) and two cross-sectional studies (Locker 1992; Tomar and Winn 1999) reported higher mean RDFS or RDS scores among smokers, but in one cohort study (Locker 1996) smoking was not found to be a significant predictor of root-surface caries in multiple logistic regression modeling.

Evidence Synthesis

Few studies have investigated the association between cigarette smoking and dental caries. The available literature is fairly consistent in suggesting that smokers may experience slightly more decayed, missing, or filled coronal tooth surfaces. In addition, smokers generally experienced more decayed or filled root surfaces than nonsmokers. However, many of the published studies did not address potential confounders of these associations. It is therefore possible that the observed associations could reflect in part the presence of other factors associated with both smoking and dental caries. Evidence for a dose-response relationship is sparse and inconsistent. Studies that examined whether quitting smoking reduced the risk of caries development were not identified.

There is little evidence for a biologic mechanism that would explain the role of smoking in the development of coronal dental caries. Methodologic considerations limit the interpretation of findings from epidemiologic studies. The few lines of investigation undertaken have been inconsistent in identifying either bacterial or salivary effects that would be expected to increase this risk.

Some evidence suggests that smoking may indirectly increase the risk for root-surface caries. The mechanism probably involves an increased exposure of root surfaces of teeth secondary to loss of periodontal attachment. This relationship may reflect the impact of smoking on periodontium and the subsequent exposure of tooth root surfaces to the oral environment.

Conclusions

1. The evidence is inadequate to infer the presence or absence of a causal relationship between smoking and coronal dental caries.

2. The evidence is suggestive but not sufficient to infer a causal relationship between smoking and root-surface caries.

Implications

To better characterize the relationship between cigarette smoking and dental caries, future investigations will need to control for potential confounding factors. These studies should be of the cohort design to allow for assessments of the effect of smoking on carious lesion formation and to determine whether smoking cessation reduces disease incidence. Investigations into an association between smoking and root-surface caries will need to apply indices that take into account the number of root surfaces at risk, such as the Root Caries Index (Katz 1980), or control for root surface exposure in trying to identify whether smoking acts through a direct or indirect mechanism.

The increased risk for root-surface caries may be due to smoking-associated periodontal destruction and subsequent exposure of root surfaces of teeth to the oral environment. Because of the causal relationship between smoking and periodontitis as well as with many other diseases, and because more than one-half of U.S. adult smokers visit a dentist each year, the dental care community has both the opportunity and the professional obligation to counsel patients who smoke to quit.

[ 2004 Health Consequences of Smoking: the Surgeon General continues next at Part  102 ]