Weitzman M, Govil N, Liu YH, Lalwani AK. Maternal Prenatal Smoking and Hearing Loss Among Adolescents. JAMA Otolaryngol Head Neck Surg. 2013;139(7):669-677. doi:10.1001/jamaoto.2013.3294
Copyright 2013 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Although smoking and secondhand smoke exposure are associated with sensorineural hearing loss (SNHL) in children and adults, the possible association between prenatal smoke exposure and hearing loss has not been investigated despite the fact that more than 12% of US children experience such prenatal exposure each year.
To investigate whether exposure to prenatal tobacco smoke is independently associated with SNHL in adolescents.
Cross-sectional data were examined for 964 adolescents aged 12 to 15 years from the National Health and Nutrition Examination Survey 2005-2006.
Participants underwent standardized audiometric testing, and serum cotinine levels and self-reports were used to identify adolescents exposed to secondhand smoke or active smokers.
Main Outcomes and Measures
Prenatal exposure was defined as an affirmative parental response to, “Did [Sample Person’s Name] biological mother smoke at any time while she was pregnant with [him/her]?” Sensorineural hearing loss was defined as an average pure-tone hearing level more than 15 dB for 0.5, 1, and 2 kHz (low frequency) and 3, 4, 6, and 8 kHz (high frequency).
Parental responses affirmed prenatal smoke exposure in 16.2% of 964 adolescents. Prenatal smoke exposure was associated with elevated pure-tone hearing thresholds at 2 and 6 kHz (P < .05), a higher rate of unilateral low-frequency SNHL (17.6% vs 7.1%; P < .05) in bivariate analyses, and a 2.6-fold increased odds of having unilateral low-frequency SNHL in multivariate analyses (95% CI, 1.1-6.4) after controlling for multiple hearing-related covariates.
Conclusions and Relevance
Prenatal smoke exposure is independently associated with higher pure-tone hearing thresholds and an almost 3-fold increase in the odds of unilateral low-frequency hearing loss among adolescents. These novel findings suggest that in utero exposure to tobacco smoke may be injurious to the auditory system.
More than 1 billion persons worldwide smoke tobacco products.1 Exposure to secondhand smoke (SHS) is a profound public health problem, with more than half of US children regularly exposed.2 Maternal smoking during pregnancy continues to be reported in approximately 12% of US pregnancies, although the rate is down from 40% in 1967 and 20% in 1989.3,4 Exposure to tobacco smoke, from in utero to adulthood, is associated with a wide variety of health problems.4,5 Prenatal smoke exposure causes obstetrical complications, such as placental abruption, placenta previa, and premature rupture of the membranes.6Quiz Ref IDApproximately 30% of small-for-gestational-age infants, 10% of preterm infants, and 5% of infant deaths are attributed to maternal smoking.6,7 For children exposed to smoke prenatally, the risk of sudden infant death syndrome is 2 to 3 times higher, and the risk of childhood asthma 1 to 2 times higher.2,8 Data also suggest that maternal smoking during pregnancy is related to childhood obesity and its metabolic complications among offspring,9,10 as well as cognitive and behavioral deficits, such as attention deficit hyperactivity disorder, decreased IQ, and learning disabilities.11
Recent studies have shown that adult smoke exposure is associated with a significant risk of hearing loss.12,13 Cruickshanks et al13 found that smokers in a population-based, cross-sectional study were 1.69 times more likely to have hearing loss than nonsmokers. Quiz Ref IDSimilarly, Agrawal et al12 found that cardiovascular risk factors, such as smoking, increased the odds of hearing loss 2-fold and involved the entire frequency spectrum from the apex to the base of the cochlea. Exposure to SHS also has been independently associated with an increased risk of hearing loss at low, middle, and high frequencies among adults.14 A recent study found an association between SHS and elevated pure-tone hearing thresholds at all frequencies, with statistically significant associations at 2, 3, and 4 kHz, for adolescents aged 12 to 19 years.15 These frequencies were consistent with the findings by Agrawal et al,12 suggesting global cochlear injury. The finding of elevated auditory thresholds in adolescents suggests that injury to the inner ear may be responsible for smoking-related hearing loss that may begin at a very early age and may occur in those who do not actively smoke but are passively exposed.
The finding of increased rates of sensorineural hearing loss (SNHL) in adolescents exposed to SHS and the more general known negative health consequences of prenatal smoke exposure were our impetus to investigate whether prenatal smoke exposure is independently associated with hearing loss among adolescents.
The National Health and Nutrition Examination Survey (NHANES) is a cross-sectional health survey that uses a complex, multistage design to collect and analyze data representative of the national, noninstitutionalized population of the United States.16 The survey was administered by the National Center for Health Statistics of the Centers for Disease Control and Prevention and approved by its institutional review board.16 All adult members of the household, as well as each adolescent subject, were interviewed at home regarding medical history, family history, current smoking status, presence of smokers in the household, medication use, and socioeconomic and demographic factors. In addition, each subject was randomly assigned to a morning, afternoon, or evening session for a physical examination and laboratory evaluation (with blood and urine samples) at a mobile health clinic.
Data from NHANES 2005-2006 were used because information about maternal smoking during pregnancy, the subjects’ SHS exposure, numerous potential confounders, and standardized audiometric testing was available for adolescents aged 12 to 15 years. Other age groups were not included because information about maternal smoking behavior during pregnancy was ascertained only from parents of children aged 15 years or younger, and audiometric data were available only for those older than 12 years. Data from NHANES 1999-2004 were not used because they include audiometric information only for participants aged 20 to 69 years.
Prenatal smoke exposure was defined as an affirmative parental response to the question: “Did [Sample Person’s Name] biological mother smoke at any time while she was pregnant with [him/her]?” Mothers who reported smoking during pregnancy were also asked whether they stopped smoking during pregnancy and, if so, when.
The NHANES evaluation includes 2 assessments of adolescent tobacco smoke exposure: cotinine levels and adolescent self-reports of smoking. Cotinine is the major metabolite of nicotine and is widely used as a biomarker for both active smoking and exposure to cigarette smoke. As in other studies,15,17 active smoking was defined as a serum cotinine level of at least 15 μg/L or self-report of smoking within the past 5 days. Exposure to SHS was defined as a detectable cotinine level less than 15 μg/L but at least 0.05 μg/L and no report of smoking within the past 5 days.17 Because cotinine levels less than 0.05 μg/L were below the detection limit, adolescents with cotinine levels in this range and without self-reports of smoking in the past 5 days were categorized as nonsmoking and unexposed to SHS.
The audiometric protocol in the study sample16 consisted of an otoscopic examination, tympanometry, and audiometry. During the audiometry sessions, air-conduction thresholds were measured for each ear at 0.5, 1, 2, 3, 4, 6, and 8 kHz across an intensity range of −10 to 120 dB. Testing was repeated at 1 kHz for each frequency; the correlation of the repeated test was 0.9 (P < .001) for both the right and left ears. The first 1-kHz test was the value used for the analysis in this study. Bone conduction thresholds were not measured for this study sample. Additional information about the methods, calibration equipment, and calibration protocol used in NHANES 2005-2006 may be found elsewhere.18
Hearing loss was considered sensorineural if the pure-tone hearing threshold was more than 15 dB, the otoscopic examination findings were normal, and the tympanogram was type A with peak admittance of at least 0.3 mL.19 In concordance with definitions of hearing loss used in other studies,20- 24 the low-frequency hearing threshold was defined as the average of pure-tone hearing levels at 0.5, 1, and 2 kHz, and the high-frequency hearing threshold as the average of pure-tone hearing levels at 3, 4, 6, and 8 kHz; low- or high-frequency SNHL was defined as a hearing threshold of more than 15 dB at these frequencies.20,25,26 Hearing loss was categorized as unilateral when the threshold average was more than 15 dB in the worse-hearing ear and bilateral when it was more than 15 dB in the better-hearing ear.
The participants’ race/ethnicity was recorded as one of the following categories: non-Hispanic black, non-Hispanic white, Mexican American, or “other,” including non-Mexican Hispanic, Asian, and Native American. The poverty to income ratio was defined as family income divided by the poverty threshold, as determined by the US Bureau of the Census for the year of the interview. Income was then classified as “poor” (poverty to income ratio ≤1) or “not poor” (poverty to income ratio >1).
Other data included potential risk factors for childhood hearing loss, such as 3 or more episodes of otitis media, allergy, and eczema,27- 32 as well as birth weight, noise exposure, and receipt of care in a neonatal intensive care unit (NICU), premature nursery, or any other special facility after birth. Information about maternal ototoxic drug use and about the adolescents’ exposure to potential infectious causes of SNHL was not available.
All statistical analyses were conducted with SAS statistical software.33 To account for the complex design of NHANES, SUDAAN statistical software34 was used to apply sampling weights to adjust for the oversampling of young children, older adults, Mexican Americans, and blacks. We used χ2 tests in bivariate analyses to test for differences in proportions, paired t tests to test for differences in means between risk factors and the different categories of hearing loss, and the Cochran-Armitage test to test for trends. Because unilateral low-frequency SNHL was significantly associated with prenatal smoke exposure in bivariate analyses (see the Results section), it was the dependent variable used in multivariate analyses. A multivariate logistic regression model was then used to test for independent associations between unilateral low-frequency SNHL and prenatal tobacco smoke exposure while controlling for socioeconomic variables, such as sex, race/ethnicity, and poverty status, as well as any other factor investigated that was significant at P < .01 in bivariate analyses. Nonsocioeconomic factors not significant at P < .10 were not included in the multivariate analyses.
The NHANES 2005-2006 sample included 1128 adolescents aged 12 to 15 years. Of the 1000 with complete audiometric and tympanometric tests, 15 had an abnormal otoscopic exemption, poor-quality tympanogram, or peak admittance response less than 0.3 mL and thus were assumed to have either conductive or mixed hearing loss and excluded from further analyses; further analyses were not performed in this cohort because it was too small. Of the remaining 985 potential subjects, 21 were excluded because information about maternal smoking during pregnancy was lacking, resulting in a final sample size of 964 for all analyses.
Quiz Ref IDOf the 964 adolescents, 155 (16.1%) had mothers who reported smoking during the pregnancy; 96 (61.9%) reported smoking during the first trimester only, 43 (27.7%) reported smoking into the second trimester, 13 (8.4%) reported smoking into the third trimester or throughout the entire pregnancy, and 3 (1.9%) either did not respond to the question or did not remember if or when they stopped smoking. Because of the small sample size in each category, the association between length of maternal smoking during pregnancy and SNHL was not subjected to statistical analysis.
Cotinine measures and adolescent self-reports of smoking status were available for 832 of the 964 subjects, of whom 388 had detectable cotinine levels less than 15 μg/L, with no report of smoking in the past 5 days, and were thus categorized as “secondhand smokers.” Another 12 subjects reported smoking in the past 5 days and were categorized as “active smokers.” Because of the small sample size of the active smoker group, the active and secondhand smoker groups were combined into a single group of 400 adolescents categorized as “SHS exposed.” Separate analyses were also performed after removing the 12 active smokers from the group. The remaining 432 adolescents were categorized as “SHS unexposed.” Because SHS exposure was a covariate and not the primary focus of these analyses, the 132 subjects with missing data on active smoking or cotinine measures were not excluded from the analyses.
Table 1 shows the association of prenatal smoke exposure and pure-tone hearing levels at various frequencies (0.5, 1, 2, 4, 5, 6, and 8 kHz) in bivariate analyses among those with or without prenatal smoke exposure. The table shows pure-tone hearing thresholds averaged across both ears and for the worse-hearing ear only. Across all frequencies, the mean pure-tone hearing level was higher in those with prenatal smoke exposure than in those who were unexposed. These elevations were statistically significant (P < .05) at 6 kHz for the average of the right and left ears and at 2 and 6 kHz for the ear with the worse hearing.
Table 2 displays the severity of SNHL for those with or without prenatal smoke exposure, broken down into the following categories used in prior studies about hearing loss,15 by pure-tone hearing threshold: normal hearing, 15 dB or less; mild hearing loss, more than 15 dB but no more than 25 dB; moderate hearing loss, more than 25 dB but no more than 40 dB; and severe hearing loss, more than 40 dB. Most hearing loss in both unexposed and exposed adolescents was mild. The general trend was for exposed adolescents to have higher rates of hearing loss at all levels of severity. In this sample, the prevalence of SNHL was as follows: unilateral low frequency, 9.5%; bilateral low frequency, 2.1%; unilateral high frequency, 15.2%; and bilateral high frequency, 4.0% (determined by summing the total percentage of hearing loss for each category >15 dB [mild, moderate, or severe hearing loss]). Of these types of SNHL, only unilateral low-frequency hearing loss was significantly associated with prenatal smoke exposure (P = .03; Cochran-Armitage trend test).
Table 3 and Table 4 show results of bivariate analyses investigating the relationship of hearing loss with a wide array of factors. The prevalence of unilateral low-frequency SNHL was significantly increased in the group with prenatal maternal smoking compared with the unexposed group (17.6% vs 7.1%:P < .05). Removing the 12 individuals who were active smokers from the group that was exposed to SHS did not alter the relationship between prenatal tobacco smoke exposure and hearing loss (data for the latter analysis not shown). Factors not significantly associated with hearing loss included SHS, NICU care, low birth weight, sex, race, poverty, and noise exposure. Because unilateral low-frequency SNHL was significantly associated with prenatal smoke exposure in bivariate analyses, it was the dependent variable used in multivariate analyses.
Table 5 shows the results of multivariate logistic regression analyses performed to investigate whether prenatal smoke exposure is independently associated with unilateral low-frequency SNHL when covariates such as NICU care, sex, race, and poverty status are controlled for. Quiz Ref IDOnly prenatal smoke exposure was significantly associated with hearing loss in this multivariate analysis. Adolescents with prenatal smoke exposure had a 2.6-fold higher odds of unilateral low-frequency SNHL than unexposed controls (95% CI, 1.1-6.4).
This study shows an association between maternal smoking during pregnancy and later hearing loss in adolescents aged 12 to 15 years. This increased risk is independent of SHS exposure, which recently has been found to be a risk factor for hearing loss among adolescents aged 12 to 19 years.15 Moreover, the frequencies associated with higher pure-tone hearing thresholds are different for prenatal and SHS exposure (2 and 6 kHz vs 2, 3, and 4 kHz),15 implying that these 2 exposure pathways may have disparate deleterious mechanisms and effects and that some of the biological injury responsible for childhood hearing loss may be traced back to the prenatal period.
The actual extent of hearing loss associated with prenatal smoke exposure in this study seems relatively modest; the largest difference in pure-tone hearing threshold between exposed and unexposed adolescents is less than 3 dB (Table 1), and most of the hearing loss is mild (Table 2). However, an almost 3-fold increased odds of unilateral hearing loss in adolescents with prenatal smoke exposure is worrisome for many reasons.
The damaging effects of prenatal smoke exposure, if the observed associations are causal, occur in many children and seem to manifest very early in life. Studies have shown that prenatal smoke exposure leads to poorer performance on auditory-related tasks, such as altered cognitive maturity and decreased responsiveness to sound stimuli, as early as the neonatal period.35- 37 Prospective studies with long-term follow-up at 12, 24, 36, and 48 months and at 9 and 12 years of age have demonstrated a dose-dependent relationship between prenatal smoke exposure and lower language and reading scores, particularly for auditory-related tasks.38- 40 However, these studies did not formally assess hearing with audiometry, and the putative mechanisms have not been uncovered. Our association of hearing loss in adolescents with prenatal smoke exposure may explain, in part or completely, the findings of poor performance on auditory-related tasks.
Prenatal smoke exposure may also set the stage for auditory injury as the adolescent ages. We believe that the unilateral hearing loss found in this study most likely represents an early stage of ear damage. We hypothesize that this unilateral hearing loss may progress to bilateral injury if the adolescent continues to be exposed to environmental smoke in the form of SHS or active smoke.
Furthermore, recent research has indicated that even mild hearing loss can affect a child’s ability to learn and communicate; children with mild hearing loss are more likely to be held back or drop out of school prematurely.19,41- 45 Therefore, the hearing loss reported in this study is only a snapshot of the potential damage prenatal smoke exposure may cause during an adolescent’s lifetime.
Hearing loss among children with in utero tobacco smoke exposure may be explained by a variety of mechanisms, including those implicated in the cause of other diseases associated with perinatal smoke exposure. One such mechanism is explained by the “thrifty phenotype hypothesis,”46 according to which fetal adaptations to in utero stressors lead to permanent endocrine and metabolic changes that ultimately cause problems in adulthood. This hypothesis has been used to explain the observed 134% higher risk for SNHL in individuals born small for gestational age.46 Prenatal maternal smoking may trigger metabolic dysfunction by causing fetal malnourishment due to altered placental architecture and function.47 Exposure to tobacco smoke in utero has been shown to cause a visible reduction in the diameter of fetal capillaries, resulting in a diminished surface area for exchange of gases and nutrients between the fetus and the mother.48 This malnutrition affects overall fetal growth, particularly the development of the central nervous system, including the auditory system.49 An alternative mechanism may be direct in utero injury and damage to the inner ear by nicotine and other chemicals in cigarette smoke.15
The results of this study fit with the finding by Agrawal et al12 of global cochlear injury in adult active smokers.50 In our sample, elevated pure-tone hearing thresholds were significantly elevated only at 2 and 6 kHz in the worse-hearing ear. However, although not statistically significant, the trend was for every other frequency tested (0.5, 1, 3, 4, 5, and 8 kHz) to have a higher pure-tone hearing threshold in subjects with in utero tobacco smoke exposure than in those without. There were similar trends for increased prevalence of bilateral low-frequency, unilateral high-frequency, and bilateral high-frequency SNHL in adolescents with a maternal history of smoking during pregnancy. This evidence suggests abnormalities in both the low and high frequencies and indicates that the entire cochlea is affected by maternal smoking during pregnancy. Quiz Ref IDAgrawal et al12 hypothesized that this global cochlear injury is due to dysfunction of the stria vascularis, reminiscent of the “strial” or “metabolic” hearing loss described by Schuknecht and Gacek.50 Although the results of our study fit with that hypothesis, further laboratory research is needed to clarify whether the stria vascularis is the precise location of global cochlear injury.
It is also notable that our data support an association between maternal smoking during pregnancy and later hearing loss, even though a substantial percentage of women reported quitting smoking sometime during the pregnancy. This implies that even brief smoke exposure during the first trimester of pregnancy has far-reaching consequences for hearing. These findings are consistent with current understanding of the first trimester as a critical period for inner ear development.51,52 Because of our sample size, we were unable to determine whether auditory function differed between those with early vs late cessation of smoking during pregnancy; this is an important question to answer with future prospective studies, because it has important public health policy implications for smoking cessation programs and counseling of pregnant women.
Our study has several limitations. The determination of prenatal smoke exposure may be error prone, because maternal smoking was identified from self-reported retrospective information. However, because individuals are more inclined to self-report a negative smoking history,53 prenatal smoke exposure is more likely to be underreported than overreported in this study. The relatively small size of this nationally representative sample probably contributed to the failure of low birth weight, NICU care, SHS, and noise exposure to be significantly associated with hearing loss. Moreover, the poverty to income ratio used to define socioeconomic status is based on data from the year of the interview, so there is no way to determine the family’s socioeconomic status at the time of the adolescent’s birth and whether it improved during the 12 to 15 years before the interview.
Another limitation is that no data were available on the amount of maternal smoking during pregnancy or other sources of SHS exposure for the fetus. The particular years of NHANES data used include no information about maternal ototoxic drug use or exposure to infectious causes of SNHL for the adolescents studied. Because these are independent risk factors for hearing loss,24 we cannot exclude the possibility of confounding variables. We also cannot determine the onset and progression of hearing loss, because we do not have results of prior hearing tests. Finally, although NHANES is a national, comprehensive survey, it is cross-sectional, thus limiting causal inferences.
Despite these limitations, the implications of the results for clinical medicine and public health may prove quite significant when the potential hearing-related morbidity associated with prenatal maternal smoking is considered. Hearing loss in the pediatric population has been estimated to affect as many as 19.5% of adolescents aged 12 to 19 years, up from 14.9% a decade ago.20,22 Again, research has shown that even mild hearing loss is associated with increased rates of social, educational, and behavioral problems in adolescents.35,36 In particular, such adolescents have increased difficulty with communication and learning. Thus, if the associations we describe are corroborated, children at risk for hearing loss because of maternal smoking during pregnancy may benefit from enhanced periodic screening with formal audiometric testing to monitor for hearing loss. Children so identified may profit from preferential seating in the classroom, enhancement of the signal-to-noise ratio with the use of sound field amplification or frequency modulation systems, speech and language therapy, and hearing aids. We hope that these findings will prompt new and more vigorous efforts to help parents, pregnant women, and their families stop smoking.
In conclusion, this study, based on a large, nationally representative sample with audiometric testing, demonstrated a significant association between prenatal smoke exposure and increased hearing loss in adolescents aged 12 to 15 years. Among those with in utero exposure to maternal smoking, the odds of unilateral low-frequency hearing loss were increased nearly 3-fold. More research is needed to determine whether hearing loss is associated with a dose-dependent increase in maternal smoking during pregnancy, identify the possible mechanisms underlying this association, and further elucidate the effects of this hearing loss on the academic, social, and cognitive performance of children and adolescents.
Corresponding Author: Michael Weitzman, MD, Department of Pediatrics, New York University School of Medicine, 550 First Ave, New York, NY 10016 (firstname.lastname@example.org).
Submitted for Publication: November 20, 2012; final revision received February 20, 2013; accepted April 15, 2013.
Published Online: June 20, 2013. doi:10.1001/jamaoto.2013.3294.
Author Contributions: Drs Weitzman and Lalwani had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition of data: Lalwani.
Analysis and interpretation of data: All authors.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Govil, Liu.
Administrative, technical, and material support: Lalwani.
Study supervision: Weitzman, Lalwani.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This study was presented in part at the Pediatric Academic Societies meeting; April 30, 2012; Boston, Massachusetts.
Correction: This article was corrected online July 23, 2013, for system-generated bracketed text in reference 18.