Cigarette Smoking and Hearing Loss: The Epidemiology of Hearing Loss Study

Context.— Clinical studies have suggested that cigarette smoking may be associated with hearing loss, a common condition affecting older adults.

Objective.— To evaluate the association between smoking and hearing loss.

Design.— Population-based, cross-sectional study.

Setting.— Community of Beaver Dam, Wis.

Participants.— Adults aged 48 to 92 years. Of 4541 eligible subjects, 3753 (83%) participated in the hearing study.

Main Outcome Measures.— The examination included otoscopy, screening tympanometry, and pure-tone air-conduction and bone-conduction audiometry. Smoking history was ascertained by self-report. Hearing loss was defined as a pure-tone average (0.5, 1, 2, and 4 kHz) greater than 25-dB hearing level in the worse ear.

Results.— After adjusting for other factors, current smokers were 1.69 times as likely to have a hearing loss as nonsmokers (95% confidence interval, 1.31-2.17). This relationship remained for those without a history of occupational noise exposure and in analyses excluding those with non–age-related hearing loss. There was weak evidence of a dose-response effect. Nonsmoking participants who lived with a smoker were more likely to have a hearing loss than those who were not exposed to a household member who smoked (odds ratio, 1.94; 95% confidence interval, 1.01-3.74).

Conclusions.— These data suggest that environmental exposures may play a role in age-related hearing loss. If longitudinal studies confirm these findings, modification of smoking habits may prevent or delay age-related declines in hearing sensitivity.

HEARING LOSS is estimated to affect 30% to 35% of adults aged 65 to 75 years in the United States, yet little is known about the etiology of this disorder.¹ Whereas hearing loss may be an inevitable consequence of aging, representing the cumulative damage from products of normal cellular metabolic processes, some studies of rural African tribes have failed to find a decline in hearing sensitivity with age.^2,3 This may suggest that genetic, environmental, and lifestyle factors play a role in the development of presbycusis, age-related hearing loss.

Cigarette smoking may affect hearing through its effects on antioxidative mechanisms or on the vasculature supplying the auditory system.^4,5 An association between cigarette smoking and hearing loss among adults has been found in some clinical studies.^6-9 Weiss⁷ found that men who smoked more than 1 pack per day had worse hearing thresholds at 250 to 1000 Hz than nonsmokers or "light" smokers, but there was no difference at higher frequencies. Siegelaub et al⁹ reported on a large study of 33146 men and women seen at Kaiser-Permanente, Oakland, Calif. Among men without a history of noise exposure, current smokers were more likely than nonsmokers to have a hearing loss at 4000 Hz, but the size of the effect was small. There was no association among women. The Baltimore Longitudinal Study of Aging¹⁰ found no association between cigarette smoking and the development of a hearing loss in 531 white, upper-middle-class men.

There have been few population-based studies of smoking and hearing. In the Health Interview Survey,¹¹ men who smoked 2 or more packs per day were more likely to report having a hearing loss than nonsmokers. In the Framingham Study,¹² which tested hearing with audiometry, there was no association between cigarette smoking and hearing loss. The purpose of our article was to evaluate the association between cigarette smoking and hearing loss in a large population-based cohort of adults aged 48 to 92 years.

A private census was conducted between September 15, 1987, and May 4, 1988, to identify residents of the city or township of Beaver Dam, Wis, who were aged 43 to 84 years.¹³ This cohort was subsequently invited to participate in the Beaver Dam Eye Study, a study of age-related ocular disorders.^14,15 Of the 5924 eligible people, 4926 (83%) participated in the eye examination phase (1988-1990). Participants of this study who were alive as of March 1, 1993, were eligible for the Epidemiology of Hearing Loss Study (n=4541), which occurred at the time of the 5-year follow-up visit for the eye study. Of those eligible, 3753 (82.6%) participated in the hearing study, 180 (4.0%) died prior to being seen, 604 (13.3%) refused to participate, and 4 (0.1%) were lost to follow-up. Some participants (n = 182) completed the interview but refused the hearing test. These participants were excluded from analyses.

The average age of participants was 65.8 years, and 57.7% were women. Comparisons between participants and nonparticipants have been reported previously.¹⁶ In brief, nonparticipants were slightly older (by an average of about 4 years), slightly more likely to be male, and more likely to have died since the examination phase began.

The hearing examination included an otoscopic evaluation,¹⁶ a screening tympanogram^17,18 (GSI 37 Autotymp, Lucas GSI Inc, Littleton, Mass), and pure-tone air-conduction and bone-conduction audiometry.¹⁶ Audiometric testing was conducted according to the guidelines of the American Speech-Language-Hearing Association¹⁹ in sound-treated booths (Industrial Acoustics Company, New York, NY) using clinical audiometers (Virtual 320, Virtual Corporation, Seattle, Wash) equipped with TDH-50 earphones (Telephonics Corp, Farmingdale, NY). Insert earphones (E-A-Rtone 3A, Cabot Safety Corp, Indianapolis, Ind) and masking were used as necessary. Pure-tone air-conduction thresholds were obtained for each ear at 250, 500, 1000, 2000, 3000, 4000, 6000, and 8000 Hz. Bone-conduction thresholds were measured at only 2 frequencies (500 and 4000 Hz) because of time constraints. Participants who were unable to travel to the clinic site (nursing home residents, homebound participants, and participants living in remote areas; n = 132) were tested at their place of residence using a portable audiometer (Beltone 112, Beltone Electronic Corp, Chicago, Ill).

All audiometers were initially calibrated in accordance with American National Standards Institute (ANSI) specifications and were recalibrated every 6 months during the study period.²⁰ Ambient noise levels were measured at each home or nursing home visit and were routinely monitored at the clinic site at the Beaver Dam Community Hospital to ensure that testing conditions complied with ANSI specifications.²¹

A questionnaire about ear and hearing-related medical history; noise exposure during leisure, military service, and work; and self-perceived hearing function was administered as an interview. Questionnaire data on medical history, lifestyle factors, and medication use were obtained as part of the 5-year follow-up examination for the Beaver Dam Eye Study (1993-1995), except for 19 people who participated in the hearing study but refused to participate in this eye examination. In these cases, the medical history and lifestyle factors interview was conducted during the hearing examination. Overall, the average time between the 5-year follow-up eye examination and the baseline hearing examination was 4.5 days. Cigarette smoking status at the time of the baseline hearing examination in 1993 through 1995 was determined by self-report. Participants were classified as either nonsmokers (ie, smoked fewer than 100 cigarettes in their lifetime), ex-smokers, or current smokers. Total pack-years smoked was defined as the number of cigarettes smoked per day divided by 20 cigarettes per pack, then multiplied by the number of years of smoking. No questions were asked about exposure to environmental tobacco smoke. As a surrogate for environmental tobacco smoke exposure in the home, nonsmoking participants were considered to be exposed if they lived in a 2-person household in which the other person was a study participant and a current smoker. Nonsmoking participants who lived either alone or in a 2-person household in which the other person was a study participant and a nonsmoker or an ex-smoker were considered unexposed. We did not have information on smoking habits for household members who were not study participants. This marker may misclassify participants exposed to tobacco smoke in other settings (eg, work or social settings) and those who have had past exposure in the home (eg, lived with a smoker in the past) as unexposed.

A history of cardiovascular disease was considered to be present if the person reported having had a stroke, myocardial infarction, or angina. Alcohol consumption was measured by a quantity and frequency questionnaire and converted to grams of ethanol per week. History of occupational noise exposure was considered positive if the person reported ever having a job where he or she had to speak in a raised voice to be heard; being a farmer and driving a tractor without a cab; or having military service with noise exposure (pilot; airplane or tank crew member; worked in the engine room of a ship; or used grenades, mortars, shoulder-held grenade launchers, or weapons systems requiring more than 1 person for operation).

For the purposes of this article, hearing loss was defined as a pure-tone average (PTA) of thresholds at 500, 1000, 2000, and 4000 Hz greater than 25-dB hearing level (dB HL) in the worse ear. The worse ear was chosen in order to include people with at least 1 affected ear. Severity of hearing loss was classified as mild (>25-dB HL and ≤40-dB HL), moderate (>40-dB HL and ≤60-dB HL), or marked (>60-dB HL) based on this pure-tone average.

Participants who reported having a hearing loss at younger than 30 years or a history of ear surgery or whose examination data showed unilateral hearing loss or hearing loss with a conductive component (air-bone gap ≥15-dB HL at 500 and/or 4000 Hz) that, if treated and resolved, would leave them with normal hearing thresholds (PTA <25-dB HL in both ears based on bone conduction) were considered to have a hearing loss that was not consistent with presbycusis. A unilateral loss was defined as one ear having a PTA of greater than 25-dB HL, the opposite ear having a PTA of 25-dB HL or less, and a 20-dB or greater difference in PTA between ears.

Analyses were conducted using SAS Version 6.09 software (SAS Institute Inc, Cary, NC). Univariate analyses used the χ² test of association for categorical variables, the Mantel-Haenszel test of trend²² for ordinal data, and Student t tests of differences in means for continuous data. Logistic regression was used to evaluate the odds of having a hearing loss associated with smoking adjusting for age, sex, and other potential confounders.²³ Interaction effects of age group and smoking and gender and smoking were tested but eliminated from the presented models as they were not statistically significant.

Table 1 presents the descriptive characteristics of participants in the study. Overall, 45.9% had a mild or greater hearing loss. Forty-six percent of participants were nonsmokers, 39.3% were ex-smokers, and 14.7% were current smokers. The number of pack-years of smoking ranged from 0 to 250, with an average of 34.9 pack-years among current smokers and 28.2 pack-years among ex-smokers. Current smokers reported smoking an average of 17.5 cigarettes per day (SD, 9.9). Smoking patterns varied greatly by age, with few older people reporting being current smokers (Table 2).

In age-specific analyses, smoking history was associated with hearing loss in all but the oldest age group (Table 2). In each age group, the prevalence of hearing loss was higher among current smokers than nonsmokers.

After adjusting for age, history of cardiovascular disease, alcohol consumption, occupational noise exposure, and education in a logistic regression model, current smokers had an increased risk of having a hearing loss compared with nonsmokers (odds ratio [OR], 1.69; 95% confidence interval [CI], 1.31-2.17) (Table 3).

Prevalent hearing loss in a cohort of older people may be caused by reasons other than presbycusis. For example, the hearing loss may have developed at a young age or may represent noise-induced hearing loss. This heterogeneity may lead to spurious associations or may make it more difficult to detect an association. To address these concerns, we conducted 2 sets of analyses to evaluate the consistency of this association in subgroups more likely to have presbycusis.

We evaluated the association between smoking and hearing loss in participants with (n=1903) and without (n= 1536) a history of occupational noise exposure because people with a history of occupational noise exposure may have noise-induced hearing loss rather than presbycusis (Table 3). The smoking and hearing loss relationship remained statistically significant in participants with a history of occupational noise exposure (OR, 1.85; 95% CI, 1.33-2.57) and those without occupational noise exposure (OR, 1.53; 95% CI, 1.03-2.29) after controlling for age, sex, history of cardiovascular disease, alcohol consumption, and education.

We excluded from the analyses participants with hearing loss that was inconsistent with presbycusis (ie, onset of hearing loss at <30 years, a history of ear surgery, a conductive hearing loss without any evidence of decreased hearing sensitivity if the conductive loss were resolved, or a unilateral hearing loss) (n=269). Current smokers, as shown in Table 3, continued to be more likely to have a hearing loss than nonsmokers (OR, 1.95; 95% CI, 1.48-2.57) after adjusting for age, sex, history of cardiovascular disease, alcohol consumption, history of occupational noise exposure, and education.

The association between pack-years of exposure and hearing loss was evaluated to determine if risk of hearing loss varied by amount of exposure. As shown in Table 4, in age-specific analyses, pack-years of smoking were associated with the prevalence of hearing loss in all but the oldest age group (P for trend <.001). After adjusting for age, sex, cardiovascular disease, education, alcohol consumption, and occupational noise exposure, pack-years of smoking remained significantly associated with hearing loss (P=.02). Those in the highest exposure category (≥40 pack-years) were 1.30 times as likely to have a hearing loss as those with 0 pack-years of exposure (95% CI, 1.04-1.63).

To test for a dose-response relationship among those exposed to cigarette smoking, these analyses were repeated excluding nonsmokers. As shown in Table 4, there was a statistically significant test of trend for the association between pack-years and hearing loss among participants aged 60 to 69 years, with a borderline association among those aged 48 to 59 years and 80 to 92 years. In a logistic regression model adjusted for age and sex, participants in the highest exposure category were 1.27 times as likely to have a hearing loss as those with fewer than 10 pack-years of smoking exposure (95% CI, 0.96-1.69; P=.09).

Among nonsmoking participants, 78.6% reported living alone or living with a study participant. In this subset (n= 1113), the first nonsmoking participant was considered the index case (the person of interest in the analysis). If the nonsmoker lived with a current smoker, the index case was considered to be exposed to environmental tobacco smoke (n= 53). Nonsmokers who lived alone or lived with a nonsmoking or ex-smoking study participant were considered currently unexposed to environmental tobacco smoke (n=1060). In this subset, those who lived with a household member who smoked were more likely to have a hearing loss than those who did not live with a smoker (OR, 1.94; 95% CI, 1.01-3.74; P=.047), after adjusting for age, sex, history of cardiovascular disease, alcohol consumption, education, and occupational noise exposure.

In this large population-based study, participants who were current cigarette smokers were 1.7 times as likely to have hearing loss as nonsmokers. This association was statistically significant after adjusting for the potential confounding effects of age, sex, education, occupational noise exposure, cardiovascular disease, and alcohol consumption. In age-specific analyses, this pattern was consistent for all but the oldest age group, which may be because of the small number of current smokers (n=13) and the effects of selective mortality. The odds of having a hearing loss increased with pack-years of smoking and were higher for nonsmokers living with a current smoker. These results are consistent with early clinical studies reporting decreased hearing sensitivity in smokers compared with nonsmokers,^4-9 animal studies showing cochlear damage after exposure to cigarette smoke,⁴ and population-based self-reported data from the Health Interview Survey.¹¹

No association between cigarette smoking and hearing loss was found in the population-based Framingham Study.¹² In that study, analyses focused on low-frequency and high-frequency hearing separately and a higher cut point (>40-dB HL) was used to classify hearing loss. These methodological differences may account, in part, for the discrepant results. For example, people with mild hearing losses and those with sloping high-frequency hearing losses typical of presbycusis were classified as having normal hearing in some analyses. This misclassification may have biased the findings, making it difficult to detect an association with smoking. In addition, analyses were conducted for men and women separately, decreasing the power to detect an association in the absence of evidence of a gender-smoking interaction. In the current study, there was no significant interaction between gender and smoking (data not shown).

The Baltimore Longitudinal Study of Aging resulted in the only article (to our knowledge) evaluating the longitudinal association between smoking and hearing loss. In that study, no association was found between current (baseline) smoking habits and the incidence of hearing loss.¹⁰ However, there were only 46 incident cases of hearing loss among the 531 men participating in the study, which may have limited the power to detect an association.

Whereas the current study relies on self-report of smoking behavior, which may be a biased measure because of underreporting, the association with number of pack-years of exposure suggests that any misclassification effects may have biased the results toward the null. There was weak evidence of a dose-response relationship among smokers after adjusting for age and sex (P=.09). The failure to detect a strong dose-response relationship may reflect the limitations of the self-reported estimate of lifetime exposure, the effects of selective mortality, or the impact of cohort effects because few participants older than 70 years had ever smoked.

The finding of an association with exposure to environmental tobacco smoke in the home is consistent with reports of effects of passive smoking on hearing sensitivity in children.^24,25 The measure of exposure to environmental tobacco smoke in the home used in this study may be a poor marker for total exposure. We did not ask questions about exposure to secondhand smoke in the home, workplace, or in social settings. Because of the population-based nature of the study, we were able to identify a subset of nonsmokers who either lived alone or with another study participant who had provided self-reported smoking behavior. However, participants currently exposed to significant amounts of environmental tobacco smoke in settings outside of the home may have been misclassified as unexposed. This crude index of exposure to environmental tobacco smoke did not include information on past exposure, intensity of exposure, or duration of exposure. Thus, these data should be interpreted cautiously. Whereas these biases would tend to increase the likelihood of failing to detect an association, it is possible that factors unaccounted for in our analytic model may explain this association. Studies using quantitative methods for measuring exposure to environmental tobacco smoke are needed to thoroughly evaluate this relationship. The point estimate for the association between environmental tobacco smoke exposure in the home and hearing loss appears to be larger than the estimate for smoking history. However, because of the broad confidence limits, this should not be interpreted as suggesting that passive smoking has a greater effect than active smoking.

Cigarette smoking is well recognized to be associated with other lifestyle and socioeconomic factors that may adversely affect health. For example, people who smoke may be more likely to be exposed to noisy workplaces and leisure activities, they may be more likely to consume alcoholic beverages, and they may have chronic conditions, such as heart disease, that may be associated with hearing loss. In models that included alcohol consumption, cardiovascular disease, education, and occupational noise exposure to adjust for potential confounding effects, smoking history remained significantly associated with hearing loss, suggesting that smoking is not merely serving as a marker for other lifestyle factors. The inclusion of leisure-time noise exposure or the number of medications used did not alter the results (data not shown).

These data suggest that cigarette smoking, a well-known risk factor for other chronic diseases, may affect hearing sensitivity. Cigarette smoking may affect hearing through its effects on antioxidative mechanisms or on the vasculature supplying the auditory system.^4,5 Recently, animal studies have identified nicotinic-like receptors in the hair cells, which suggests that smoking may have direct ototoxic effects on hair cell function through its potential effect on the neurotransmission of auditory stimuli.^26,27

In summary, the findings from this cross-sectional, population-based study are consistent with the literature and suggest that environmental exposures may play a role in age-related hearing loss. If longitudinal, population-based studies confirm these findings, modification of smoking habits may prevent or delay age-related declines in hearing sensitivity.