Objective
To estimate the relative risk for otitis media (OM) in children from environmental tobacco smoke (passive exposure), maternal smoking during pregnancy (gestational exposure), or both.
Design
Analysis of data from a national cross-sectional health survey, utilizing questionnaire information and serum cotinine measurements.
Participants
Children younger than 12 years (N = 11 728) in the Third National Health and Nutrition Examination Survey (NHANES III), conducted from 1988-1994.
Main Outcome Measures
Occurrence and recurrence of ear infections.
Results
The cumulative incidence of ear infections was 69%. Of all participants, 38% were exposed to passive smoke, 23% were exposed to gestational smoke, and 19% were exposed to combined passive and gestational smoke. The occurrence of any ear infection was not increased by passive smoke exposure (adjusted risk ratio [RR], 1.01; 95% confidence interval [CI], 0.95-1.06), but was slightly increased by gestational (adjusted RR, 1.08; 95% CI, 1.01-1.14) and combined (adjusted RR, 1.07; 95% CI, 1.00-1.14) smoke exposures. The risk of recurrent ear infections (≥6 lifetime episodes) was significantly increased with combined smoke exposure (adjusted RR, 1.44; 95% CI, 1.11-1.81). Other risk factors for ear infection identified in multivariable analysis were race/ethnicity, poverty-income ratio of 2.00 or more, attendance in day care, history of asthma, and presence of allergic symptoms.
Conclusions
Passive smoke exposure was not associated with an increased risk of ever developing an ear infection in this study. The increased risk found with gestational and combined smoke exposures has marginal clinical significance. For recurrent ear infections, however, combined smoke exposure had a clinically and statistically significant effect.
OTITIS MEDIA (OM) is a common complication of upper respiratory tract infections in young children.1 Among the risk factors often associated with recurrent OM are a family history of OM, child care outside the home, presence of siblings, early onset of OM, atopy, male sex, not being breastfed, and parental smoking.1-13
Exposure to environmental tobacco smoke (ETS), often called passive smoking, is thought to increase the risk of OM, possibly through an effect on mucociliary clearance. Agius et al2 found that in patients who had chronic OM with effusion, ciliary beat function was impaired with, but not without, passive smoke exposure. Other possible mechanisms include goblet cell hyperplasia with mucus hypersecretion and alteration of phagocytic antibacterial defenses in conjunction with some viral infections.14 The role of ETS in children's middle ear disease (eg, acute OM, recurrent OM, or OM with effusion) has been clinically investigated in numerous cohort studies,3,5,9,14-23 case-control studies,7,10,24-36 and cross-sectional surveys8,12,37-45 and has been summarized in multiple reviews46,47 and meta-analyses.4,48,49 Quantitative synthesis of results in the 3 meta-analyses suggests a small to moderate risk increase from parental smoking, with estimates ranging from a risk ratio (RR) of 1.19 for middle ear disease to an odds ratio (OR) of 1.74 for recurrent OM (≥3 episodes).
Of the studies that do show a link between ETS and OM, several have suggested that maternal, but not paternal, smoking increases the risk of OM.5,12,16,17,29,50 Of 4 studies that have investigated the association between maternal smoking during pregnancy and OM, 2 have suggested that the incidence of OM increased with gestational smoke exposure, as distinguished from maternal smoking postpartum.8,12,17,50
Maternal smoking during pregnancy is generally known to be associated with increased adverse neonatal events51-53 and increased childhood morbidity, such as depressed neonatal respiratory function and increased incidence of childhood asthma.54-63 Gestational smoking is also associated with cord blood immune function alterations in newborns, including reduction in natural killer cell activity,64 decreased neutrophil counts,65 and elevated levels of IgA, IgM, and IgG.66 Because OM is considered a complication of an upper respiratory tract infection, it is pathophysiologically plausible that maternal smoking during pregnancy would increase the risk of OM. The current study was conducted to answer the question: Does maternal smoking during pregnancy and/or subsequent passive smoke exposure raise the risk of OM in children?
We used data from the Third National Health and Nutrition Examination Survey (NHANES III), a national cross-sectional health survey performed from 1988-1994. The survey included 33 994 persons aged 2 months and older who represented the noninstitutionalized civilian population of the United States. Details of the complex sampling design, data collection, and weighting approach have been described elsewhere.67 Briefly, NHANES III used a stratified, multistage probability sample design, with oversampling of young children (age <5 years), older persons (age >59 years), non-Hispanic black persons, and Mexican Americans. To obtain a distribution of participants similar to the US population as a whole, sampling weights were applied during the analysis to incorporate the differential probabilities of selection and include adjustments for noncoverage and nonresponse.67 To generate estimates of risk for OM in children, the current analysis is limited to the 11 728 survey participants younger than 12 years.
In the NHANES III interviews on demographic characteristics, the family reference person (ie, head of household) marked race/ethnicity as non-Hispanic white, non-Hispanic black, or Mexican American. Any person who did not choose a category was considered to be "other." A poverty-income ratio (PIR), calculated as the reported household annual income divided by the poverty threshold defined by the US Census Bureau with adjustment for family size, was coded as a continuous variable in the NHANES III database. For our analysis, the PIR was demarcated into 3 ordinal categories: less than 1.00, 1.00 to 1.99, and 2.00 or more. Educational level completed by the family reference person was coded in the NHANES III database as "never attended school or kindergarten only" or in individual levels of 1 to 17 years, with 17 entered for those with 17 or more years of schooling. For the current analysis, educational level was categorized as less than 7 years (primary or no schooling), 7 to 12 years (at least some secondary school), and more than 12 years (at least some college).
For each child, all information was given by a proxy respondent (ie, mother, father, sibling, grandparent, aunt or uncle, or other). Age of the child was recorded in the survey in months or years and analyzed in the current study by year. Day care attendance was defined in the survey as ever having attended a day care center with at least 6 children and whether attendance was less than 10 or at least 10 hours per week. Child's birth weight was divided into 3 ordinal categories: less than 2500 g, 2500 to 4100 g, and more than 4100 g. For children younger than 6 years, questions were asked about whether breastfeeding had ever occurred, and if so, the child's age when it completely stopped.
For the question, "Did [the child] ever have an ear infection or an earache?" a positive response was followed by questions about whether a physician ever treated the child for an ear infection; whether tubes were placed for treatment; and the number of episodes, coded in the survey as 1, 2, 3 to 5, or 6 or more episodes. Any positive response was counted as an ear infection, and we defined recurrent ear infections as at least 6 episodes18 and nonrecurrent as 1 to 5 episodes. Because other investigators have considered 3 or more ear infections to be recurrent, an analysis using this definition of recurrence was also done separately.
The presence of comorbid conditions—asthma, chronic bronchitis, and hay fever—was ascertained by the question, "Did a doctor ever say that [the child] had . . . ?" Presence of allergic symptoms was elicited by the questions, "During the past 12 months, has [the child] had any episodes of stuffy, itchy, or runny nose? Watery, itchy eyes?"
Passive cigarette smoke exposure in the NHANES III survey was determined by questions about the existence and number of persons who currently smoked cigarettes in the household. We coded any passive exposure as yes or no for the presence of smokers in the household and quantified it by summing the number of smokers and the number of cigarettes they smoked per day. Maternal smoking during pregnancy (ie, gestational exposure) was ascertained in the survey by asking whether the mother smoked at any time while pregnant with the child. A positive response was followed by a question about whether she quit or refrained from smoking for the rest of the pregnancy. We then created a composite variable for the combination of passive smoking and gestational smoke exposure. For this composite variable, exposure was categorized as neither passive nor gestational, passive only, gestational only, or both passive and gestational (ie, combined).
To quantify the dose of passive cigarette smoke exposure in another way, we examined the NHANES III results for children's values of serum cotinine, a metabolite of nicotine. Because this test was performed only on children who participated from 1988-1991, serum cotinine measurements were available for only 60% of the total survey population. Furthermore, because serum cotinine was not measured in children younger than 4 years, 5982 children were excluded from the serum cotinine analysis. Data from 1825 children were thus used for the analysis of serum cotinine. Concentrations were determined by the NHANES III survey in a 2-step process. The enzyme immunoassay method was used as a screening method for differentiating "low" (<25 ng/mL) and "high" (≥25 ng/mL) cotinine concentrations. Based on the low or high concentrations on enzyme immunoassay, liquid chromatography/mass spectrometry analysis was performed on all specimens. Serum cotinine concentrations greater than 10 to 20 ng/mL generally indicate active smokers.68
All analyses for this study were done using SUDAAN software (Research Triangle Institute, Research Triangle Park, NC) to incorporate sampling weights (ie, primary sampling unit and strata information) consistent with the complex design of the NHANES III survey.67 This method of statistical adjustment, incorporating the sampling weights, produced a weighted cumulative incidence that estimates the proportion of children who ever had ear infections nationally. Bivariate comparisons of the cumulative incidence of ear infections were assessed using χ2 tests and crude RRs with 95% confidence intervals (CIs) for dichotomous independent variables. The logistic regressions produced ORs and 95% CIs for categorical independent variables after they were converted to "dummy" binary variables. For continuous serum cotinine level, medians and interquartile ranges were used.
A second method of statistical adjustment, a multivariable adjustment, controlled for the effect of multiple independent variables. Demographic, baseline, and comorbidity variables were used in the multivariable logistic regression models. For OR calculations in the logistic regression models, the reference groups were non-Hispanic whites for the race/ethnicity variable, female sex, and normal birth weight (ie, 2500-4100 g). The adjusted ORs obtained from logistic regression were converted into adjusted RRs for ear infections (any, recurrent, and nonrecurrent) because they all had cumulative incidences of more than 10% in this population. The conversion, performed with an equation provided by Zhang and Yu,69 was done because the OR for relatively common outcomes will either overestimate RRs more than 1 or underestimate those less than 1 (ie, provide estimates farther away from 1 than the true RR). The 95% CIs for adjusted RRs were correspondingly converted from ORs using the same equation. Multiple stratification tables were used to examine possible interactions between the independent variables in affecting the cumulative incidence of any ear infections.70
The US cumulative incidence of any ear infection in children younger than 12 years was 69.1%, affecting similar proportions of men and women (Table 1). The rates of ear infection rose from 34.1% among those younger than 1 year to a maximum of 77.2% among 5-year-olds before decreasing to 69.5% among 11-year-olds (data not shown). In bivariate analysis, the risk of having any ear infection was lower in non-Hispanic blacks, Mexican Americans, and those of other race than in non-Hispanic whites. Both increasing PIR and level of educational attainment by the head of household increased the risk for any ear infection in children. The risk also increased, as expected, with attendance in day care but was not affected by whether the child attended day care less than 10 or 10 or more hours per week (data not shown). Birth weight less than 2500 g was associated with a decreased rate of ear infections, but breastfeeding for longer than 4 months was not. In addition, history of asthma and allergic symptoms were significantly associated with an increased cumulative incidence of any ear infections, although chronic bronchitis and hay fever were not. Of the variables significantly associated with any ear infections in bivariate analysis, only educational level of head of household did not continue to be a significant factor in multivariable analysis (data not shown).
Any passive smoke exposure occurred in 38%, any gestational smoke exposure in 23%, and combined passive and gestational smoke exposure in 19% of children younger than 12 years. Although any passive smoke exposure at home had no statistically significant association with an increased risk of any ear infection in children (adjusted RR, 1.01; 95% CI, 0.95-1.06), any maternal smoking during pregnancy did confer a small increased risk for developing any ear infections (adjusted RR, 1.08; 95% CI, 1.02-1.14) (Table 2). Whether the mother quit smoking during pregnancy did not significantly alter the cumulative incidence of any ear infections in bivariate analysis (74.5% for children whose mothers did quit vs 73.6% for those whose mothers continued to smoke). Defining any ear infections as only those treated by physicians decreased the cumulative incidence to 66.9% (data not shown). This definition, however, did not change the risk of exposure to passive smoke (adjusted RR, 1.01; 95% CI, 1.95-1.06) or maternal smoking during pregnancy (adjusted RR, 1.08; 95% CI, 1.01-1.15).
Dose of passive smoke exposure
The dose of smoke exposure was quantified according to the number of persons smoking at home, the number of cigarettes smoked at home, and the composite variable of passive and gestational exposures (Table 3). The median serum cotinine levels rose with each increasing ordinal category of passive smoke exposure, confirming the accuracy of statements regarding the dose exposures. The occurrence of any ear infections was not related to the number of persons (χ2 test for linear trend, 0.29; P<.60) or number of cigarettes smoked at home (χ2 test for linear trend, 1.06; P<.40), but there was a slight statistically significant increase in cumulative incidence (adjusted RR, 1.07; 95% CI, 1.00-1.14) for combined exposure.
When the dose of passive smoke exposure was quantified according to serum cotinine levels, no significant dose response was found in the cumulative incidence of any ear infections (χ2 test for linear trend, 1.15; P<.30) (Table 4). Although the unadjusted RR for those with a serum cotinine level less than 3 to 4 ng/mL was 1.14, the 95% CI included 1, and the RRs for higher serum cotinine levels decreased rather than increased. Only the unadjusted RRs are given to avoid numerical instability from too few denominator degrees of freedom for the full multivariable logistic regression model.71
We analyzed the potential effect of comorbid conditions (asthma, chronic bronchitis, hay fever, or allergic symptoms), passive smoke exposure, and the cumulative incidence of any ear infections in multiple stratification tables (data not shown). Within subgroups of children with these comorbid conditions, however, passive smoke did not significantly change the cumulative incidence of any ear infections.
Table 5 presents the effect of tobacco exposure on nonrecurrent vs recurrent ear infections. Their cumulative incidence was not significantly altered by any passive smoking alone, but any maternal smoking during pregnancy increased the risk of recurrent over nonrecurrent ear infections. Furthermore, in the analysis of the composite exposure variable, only the presence of both passive and gestational smoke exposure produced a significantly increased risk of recurrent ear infection (adjusted RR, 1.41; 95% CI, 1.09-1.78). Expressed as the number needed for 1 extra effect (calculated as the reciprocal of the absolute difference in cumulative incidence between those who were and were not exposed to both passive and gestational smoke),72 21 children must be so exposed to result in 1 excess case of recurrent ear infections. However, within the subgroup of children with recurrent ear infections, exposure to passive smoke (crude RR, 0.97; 95% CI, 0.73-1.28), maternal smoking during pregnancy (crude RR, 1.09; 95% CI, 0.78-1.53), or both (crude RR, 1.08; 95% CI, 0.73-1.57) did not result in higher rates of tympanostomy tube placement than no exposure. Only the unadjusted RRs are given here for the tympanostomy tube placement data to avoid numerical instability from too few denominator degrees of freedom for the full multivariable logistic regression model.71
If the definition of recurrent was 3 or more ear infections, passive smoking still did not significantly raise the risk of recurrent ear infections (adjusted RR, 1.03; 95% CI, 0.90-1.24). Exposure to maternal smoking during pregnancy or combined exposure continued to significantly increase the risk of recurrent ear infections (adjusted RR, 1.19; 95% CI, 1.03-1.35; and adjusted RR, 1.18; 95% CI, 1.01-1.36, respectively) with this definition of recurrence.
Other risk factors associated with recurrent ear infections differed slightly from the risk factors associated with any ear infections. The risk of recurrent ear infections continued to be lower among non-Hispanic blacks (adjusted RR, 0.28; 95% CI, 0.22-0.36), Mexican Americans (adjusted RR, 0.52; 95% CI, 0.43-0.63), and those of other race (adjusted RR, 0.59; 95% CI, 0.34-0.98) than in non-Hispanic whites; higher among those with a PIR of 2.00 or more (adjusted RR, 1.76; 95% CI, 1.32-2.28) than those with a PIR less than 1.00; higher in those attending day care (adjusted RR, 1.62; 95% CI, 1.31-1.96); and higher in those with asthma (adjusted RR, 1.64; 95% CI, 1.32-2.09) or allergic symptoms (adjusted RR, 1.68; 95% CI, 1.32-2.09 for nasal or ocular symptoms; and adjusted RR, 2.19; 95% CI, 1.78-2.65 for nasal and ocular symptoms). Educational level of the head of household, birth weight, and breastfeeding duration were not significant factors in multivariable analysis, but male sex had a small effect (adjusted RR, 1.23; 95% CI, 1.02-1.47).
Our results suggest that maternal smoking during pregnancy may be a risk factor for ear infections and that the combination of gestational and passive smoke exposure moderately increases the risk of recurrent ear infections. Prior studies examining the effect of maternal smoking during pregnancy vs ETS have reported conflicting results. In a cohort study that followed 5627 women from their first prenatal visit to 5 years postdelivery, Stathis et al50 found that any amount of maternal smoking at the first prenatal visit was associated with increased risk of acute ear infections at 5 years postdelivery (1-9 cigarettes/d: OR, 1.6; 95% CI, 1.1-2.5; 10-19 cigarettes/d: OR, 2.6; 95% CI, 1.6-4.2; and 20+ cigarettes/d: OR, 3.3; 95% CI, 1.9-5.9). Smoking at the first prenatal visit was also associated with increased subacute infections (1-9 cigarettes per day: OR, 1.7; 95% CI, 1.0-3.0; 10-19 cigarettes per day: OR, 2.6; 95% CI, 1.4-5.0; and 20+ cigarettes per day: OR, 2.8; 95% CI, 1.3-6.0), and smoking 20 or more cigarettes per day was associated with increased risk of ear surgery (OR, 2.9; 95% CI, 1.3-6.6). Smoking during the third trimester and at 6 months and 5 years postdelivery, however, did not increase the risk of acute OM, subacute ear infections, or ear surgery. Ey et al17 followed up 1013 infants from birth to 1 year of age and found that compared with children whose mothers did not smoke, recurrent OM occurred more frequently in children whose mothers were heavy smokers both during pregnancy and after delivery (32%), but not when mothers smoked only during pregnancy (15%). In a cross-sectional study of 2065 children aged 2½ to 6 years, maternal smoking during the first 3 months of pregnancy increased the proportion of children who had acute OM (RR, 1.16; P<.05).12 Smoking during the last 6 months of pregnancy also significantly raised the proportion of children with acute OM, although the difference was "less striking." In a study that used data from 2 cross-sectional surveys done in 1981 and 1988, recurrent OM occurred in 23.8% of children whose mothers smoked during pregnancy and 22.7% of those whose mothers did not (OR, 1.1; 95% CI, 0.9-1.2).8
The current result for the effect of passive smoke exposure on any ear infections is not consistent with the results of the 3 meta-analyses synthesizing studies on passive smoking and OM.4,48,49 Several explanations may account for this difference. One is that the true effect of passive smoking on ear infections (if any exists) is probably small (ie, RR <1.5), and uncontrolled confounding in this study may have moved the estimated RR from a theoretical 1.25 to 1.01 in the current analysis. Most children have an ear infection at some point, so looking at the cumulative incidence of any ear infection may ultimately result in a RR near 1.0 for risk factors that do not have a large impact. Another possible explanation is that the risk attributed to parental smoking found in previous studies may actually be driven by the combined effect of maternal smoking during pregnancy and ETS during childhood, as found in the current study, but not separated out. The magnitude of the effect of combined gestational and passive exposure in this study (adjusted RR, 1.41) is in the same range as the results of the 3 meta-analyses. Those results showed small to moderate increased risk from parental smoking, ranging from a RR of 1.19 for middle ear disease to an OR of 1.74 for recurrent OM (ie, ≥3 episodes).4,48,49 Because ORs poorly estimate the true RR when the prevalence or incidence of an outcome exceeds 10%, the OR of 1.74 can be converted69 to an RR of 1.42 to 1.51 (for an estimated recurrent OM cumulative incidence in the range of 20%-30%).
A limitation of this study is that ear infections are reported by proxy in a questionnaire in the NHANES III survey, rather than documented by a physician. Consequently, earache without ear infection, otitis externa, and ear-pulling behavior in an infant may be included in proxy-reported "ear infections." The cumulative incidence of "doctor-treated" ear infections in NHANES III, however, was not significantly different from self-report (66% for any "doctor-treated" ear infections vs 69% for self-report). We chose not to use the cumulative incidence of "doctor-treated" ear infections because the questions regarding the numbers of episodes of ear infections, which were used to define recurrence, were based on self-report. In addition, when the National Health Interview Survey used the same method of proxy-report, the 12-month cumulative incidence of acute ear infections in 1994 was found to be 62.7% in those younger than 5 years,73 a result appropriately less than the 69.1% noted in our study for ever having ear infections.
Other factors that may affect the accuracy of questionnaire-reported ear infections include the number of infections, the duration of recall, and the characteristics of the respondents. Daly et al74 found that compared with medical records, parents tended to overestimate the number of episodes if they reported 6 or more episodes and underestimate them if they reported fewer episodes. Pless and Pless75 showed that younger parents tended to recall more accurately than older parents and mothers were more accurate than fathers. Alho76 found that questionnaire-based data tended to damp the associations between various risk factors and acute OM. These same studies, however, have found good to excellent concordance between parental report and medical records for OM, suggesting that parental report is a reasonably valid approach.74-76 Also, when the analysis in the current study was repeated in children younger than 3 years, a subgroup in whom recall bias should be less, the results for passive smoking (adjusted RR, 1.03; 95% CI, 0.97-1.09) and gestational smoking (adjusted RR, 1.08; 95% CI, 1.02-1.14) are similar to those for the entire population.
Other limitations of the NHANES III data are that smoking behavior by caregivers may be underreported and that only household exposures are noted for ETS, without recognition of exposures in other settings (eg, a private baby-sitter). Underreporting of smoke exposure may bias the analysis of the effect of ETS on ear infections toward the null. A recent report using NHANES III data examined the possible discrepancy between self-reported cigarette smoking status and serum cotinine levels in adults, however, and concluded that "self-reported smoking status among adult respondents to a population-based survey conducted in a private medical setting is accurate."77 Because this current study used the same data source as that report, bias from underreport likely has minimal effect on our results. The NHANES III questionnaire does not differentiate among maternal, paternal, and other sources of ETS within the home. Nevertheless, in Table 3, the children's median serum cotinine levels, a physiologic measurement, show appropriately corresponding increases with the reported doses of smoke exposure according to the number of persons who smoke and the cumulative number of cigarettes. Therefore, ETS exposure inside the home significantly influences serum cotinine levels in children and probably accounts for most of the smoke exposure.
Additional limitations of this study come from the cross-sectional design of the NHANES III survey. Although gestational smoke exposure necessarily came before the onset of ear infections, ETS was ascertained at the time of the survey rather than before the onset of infections. A potential association between ETS and ear infections may be biased toward the null if children with ear infections came from more households with ETS exposure in the past and/or children without ear infections came from more households without ETS in the past. Similarly, ear infection status was ascertained at the time of the survey rather than at the time of infection, so that bias toward the null could occur if fewer caregivers from smoking households recalled children having ear infections than from nonsmoking households.
The socioeconomic and demographic risk factors in this study are similar to those in another national cross-sectional survey of OM.8 In particular, rates of ear infections were inversely correlated with non-Hispanic black and Mexican American race/ethnicity, but directly correlated with increasing income levels. Two other studies have found elevated rates of OM with increasing income16 and socioeconomic status.5 In contrast, however, another study found that rates of OM or middle-ear effusion were increased by low socioeconomic status and black race.9 The small statistically significant increase of OM in the male sex has been documented in multiple reports.3,5,6,8,11,16,17,36
Other important risk factors for recurrent ear infections in this study include attendance in day care, which is almost universally found in other studies of OM, and the presence of allergy or allergic symptoms, which is usually found if these symptoms are studied. Breastfeeding, however, was not a protective factor in this study, a phenomenon found in many studies5,7,8,11,13,16,18,21,22,25,35,36,39,41,44 but not in others that are often quoted.3,6,9,17,31 The heterogeneity of the sampled population in the NHANES III survey allowed multiple socioeconomic and demographic factors to be analyzed simultaneously and may account for the differences found in these factors.5,7,8,11,13,16,18,21,22,25,35,36,39,41,44
In conclusion, passive smoke exposure was not associated with an increased risk of children ever experiencing ear infections in this study. The increased risk found with gestational and combined smoke exposures has marginal clinical significance. For recurrent ear infections, however, combined gestational and passive smoke exposure had a clinically and statistically significant effect, which may partially explain why maternal but not paternal smoking has been found in some studies to increase the cumulative incidence of OM. The results suggest that attempts to decrease the recurrence of OM should consider interventions against maternal smoking during and after pregnancy.
Accepted for publication October 25, 2001.
This study was presented at the Robert Wood Johnson Clinical Scholars Program National Meeting, Fort Lauderdale, Fla, November 9, 2000.
Although ETS exposure has been associated with a small increased risk of OM in children, prior studies examining the effect of maternal smoking during pregnancy vs ETS on the risk of ear infections in children have reported conflicting results. These studies have not been able to determine the effect of both of types of passive smoke exposures separately and in combination.
The study found that for recurrent ear infections, combined exposure to ETS and maternal smoking during pregnancy had a clinically and statistically significant effect. However, ETS was not associated with an increased risk of children ever experiencing ear infections in this study, and the increased risk found with maternal smoking during pregnancy and combined smoke exposures had marginal clinical significance. The results suggest that attempts to decrease the recurrence of otitis media should consider interventions against maternal smoking both during and after pregnancy.
Corresponding author and reprints: Judith E. C. Lieu, MD, Department of Otolaryngology–Head and Neck Surgery, Washington University School of Medicine, 1 Children's Place, Room 3S35, St Louis, MO 63110 (e-mail: lieuj@msnotes.wustl.edu).
1.Wald
ERGuerra
NByers
C Upper respiratory tract infections in young children: duration of and frequency of complications.
Pediatrics. 1991;87129- 133
Google Scholar 2.Agius
AMWake
MPahor
ALSmallman
LA Nasal and middle ear ciliary beat frequency in chronic suppurative otitis media.
Clin Otolaryngol. 1995;20470- 474
Google ScholarCrossref 3.Teele
DWKlein
JORosner
B Epidemiology of otitis media during the first 7 years of life in children in greater Boston.
J Infect Dis. 1989;16083- 94
Google ScholarCrossref 4.Uhari
MMantysaari
KNiemela
M A meta-analytic review of the risk factors for acute otitis media.
Clin Infect Dis. 1996;221079- 1083
Google ScholarCrossref 5.Bennett
KEHaggard
MP Accumulation of factors influencing children's middle ear disease.
J Epidemiol Community Health. 1998;52786- 793
Google ScholarCrossref 6.Alho
OPLaara
EOja
H How should relative risk estimates for acute otitis media in children aged less than 2 years be perceived?
J Clin Epidemiol. 1996;499- 14
Google ScholarCrossref 7.Kraemer
MJRichardson
MAWeiss
NS
et al. Risk factors for persistent middle-ear effusions: otitis media, catarrh, cigarette smoke exposure, and atopy.
JAMA. 1983;2491022- 1025
Google ScholarCrossref 8.Lanphear
BPByrd
RSAuinger
PHall
CB Increasing prevalence of recurrent otitis media among children in the United States.
Pediatrics. 1997;99E1
Google ScholarCrossref 9.Paradise
JLRockette
HEColborn
DK
et al. Otitis media in 2253 Pittsburgh-area infants.
Pediatrics. 1997;99318- 333
Google ScholarCrossref 10.Pukander
J Acute otitis media among rural children in Finland.
Int J Pediatr Otorhinolaryngol. 1982;4325- 332
Google ScholarCrossref 11.Rovers
MMZielhuis
GAStraatman
HIngels
Kvan der Wilt
GJvan den Broek
P Prognostic factors for persistent otitis media with effusion in infants.
Arch Otolaryngol Head Neck Surg. 1999;1251203- 1207
Google ScholarCrossref 12.Van Cauwenberge
PB Relevant and irrelevant predisposing factors in secretory otitis media.
Acta Otolaryngol Suppl. 1984;414147- 153
Google ScholarCrossref 13.Zielhuis
GAHeuvelmans-Heinen
EWRach
GHvan den Broek
P Environmental risk factors for otitis media with effusion in preschool children.
Scand J Prim Health Care. 1989;733- 38
Google ScholarCrossref 14.Etzel
RAPattishall
ENHaley
NJFletcher
RHHenderson
FW Passive smoking and middle ear effusion among children in day care.
Pediatrics. 1992;90228- 232
Google Scholar 15.Alho
OPLaara
EOja
H Public health impact of various risk factors for acute otitis media in northern Finland.
Am J Epidemiol. 1996;1431149- 1156
Google ScholarCrossref 16.Collet
JPLarson
CPBoivin
JFSuissa
SPless
IB Parental smoking and risk of otitis media in preschool children.
Can J Public Health. 1995;86269- 273
Google Scholar 17.Ey
JLHolberg
CJAldous
MBWright
ALMartinez
FDTaussig
LM Passive smoke exposure and otitis media in the first year of life. Group Health Medical Associates.
Pediatrics. 1995;95670- 677
Google Scholar 18.Harsten
GPrellner
KHeldrup
JKalm
OKornfalt
R Recurrent acute otitis media: a prospective study of children during the first 3 years of life.
Acta Otolaryngol. 1989;107111- 119
Google ScholarCrossref 19.Maw
ARBawden
R Factors affecting resolution of otitis media with effusion in children.
Clin Otolaryngol. 1994;19125- 130
Google ScholarCrossref 20.Rasmussen
F Protracted secretory otitis media: the impact of familial factors and day care center attendance.
Int J Pediatr Otorhinolaryngol. 1993;2629- 37
Google ScholarCrossref 21.Sipila
MKarma
PPukander
JTimonen
MKataja
M The Bayesian approach to the evaluation of risk factors in acute and recurrent acute otitis media.
Acta Otolaryngol. 1988;10694- 101
Google ScholarCrossref 22.Tainio
VMSavilahti
ESalmenpera
LArjomaa
PSiimes
MAPerheentupa
J Risk factors for infantile recurrent otitis media.
Pediatr Res. 1988;23509- 512
Google ScholarCrossref 23.Zielhuis
GAGerritsen
AAGorissen
WH
et al. Hearing deficits at school age: the predictive value of otitis media in infants.
Int J Pediatr Otorhinolaryngol. 1998;44227- 234
Google ScholarCrossref 25.Black
N The aetiology of glue ear: a case-control study.
Int J Pediatr Otorhinolaryngol. 1985;9121- 133
Google ScholarCrossref 26.Green
RECooper
NK Passive smoking and middle ear effusions in children of British servicemen in West Germany.
J R Army Med Corps. 1991;13731- 33
Google ScholarCrossref 27.Hinton
AEBuckley
G Parental smoking and middle ear effusions in children.
J Laryngol Otol. 1988;102992- 996
Google ScholarCrossref 28.Hinton
AE Surgery for otitis media with effusion in children and its relationship to parental smoking.
J Laryngol Otol. 1989;103559- 561
Google ScholarCrossref 29.Ilicali
OCKeles
NDeger
KSavas
I Relationship of passive cigarette smoking to otitis media.
Arch Otolaryngol Head Neck Surg. 1999;125758- 762
Google ScholarCrossref 30.Kitchens
GG Relationship of environmental tobacco smoke to otitis media in young children.
Laryngoscope. 1995;1051- 13
Google ScholarCrossref 31.Pukander
JLuotonen
JTimonen
MKarma
P Risk factors affecting the occurrence of acute otitis media among 2- to 3-year-old urban children.
Acta Otolaryngol. 1985;100260- 265
Google ScholarCrossref 32.Rovers
MMHofstad
EAFranken–van den Brand
KI
et al. Prognostic factors for otitis media with effusion in infants.
Clin Otolaryngol. 1998;23543- 546
Google ScholarCrossref 33.Rowe-Jones
JMBrockbank
MJ Parental smoking and persistent otitis media with effusion in children.
Int J Pediatr Otorhinolaryngol. 1992;2419- 24
Google ScholarCrossref 34.Stenstrom
CIngvarsson
L Otitis-prone children and controls: a study of possible predisposing factors, 2: physical findings, frequency of illness, allergy, day care, and parental smoking.
Acta Otolaryngol. 1997;117696- 703
Google ScholarCrossref 35.Stenstrom
RBernard
PABen-Simhon
H Exposure to environmental tobacco smoke as a risk factor for recurrent acute otitis media in children under the age of 5 years.
Int J Pediatr Otorhinolaryngol. 1993;27127- 136
Google ScholarCrossref 36.Takasaka
T Incidence, prevalence, and natural history of otitis media in different geographic areas and populations: epidemiology of otitis media with effusion in Japan.
Ann Otol Rhinol Laryngol. 1990;9913- 14
Google Scholar 37.Chayarpham
SStuart
JChongsuvivatwong
VChinpairoj
SLim
A A study of the prevalence of and risk factors for ear diseases and hearing loss in primary school children in Hat Yai, Thailand.
J Med Assoc Thai. 1996;79468- 472
Google Scholar 38.Iversen
MBirch
LLundqvist
GRElbrond
O Middle ear effusion in children and the indoor environment.
Arch Environ Health. 1985;4074- 79
Google ScholarCrossref 39.Lubianca Neto
JFBurns
AGLu
LMombach
RSaffer
M Passive smoking and nonrecurrent acute otitis media in children.
Otolaryngol Head Neck Surg. 1999;121805- 808
Google Scholar 40.Porro
ECalamita
PRana
IMontini
LCriscione
S Atopy and environmental factors in upper respiratory infections: an epidemiological survey on 2304 schoolchildren.
Int J Pediatr Otorhinolaryngol. 1992;24111- 120
Google ScholarCrossref 41.Reed
BDLutz
LJ Household smoking exposure: association with middle ear effusions.
Fam Med. 1988;20426- 430
Google Scholar 42.Strachan
DPJarvis
MJFeyerabend
C Passive smoking, salivary cotinine concentrations, and middle ear effusion in 7-year-old children.
BMJ. 1989;2981549- 1552
Google ScholarCrossref 43.Strachan
DP Impedance tympanometry and the home environment in 7-year-old children.
J Laryngol Otol. 1990;1044- 8
Google ScholarCrossref 44.Vinther
BElbrond
OPedersen
CB A population study of otitis media in childhood.
Acta Otolaryngol Suppl. 1979;360135- 137
Google Scholar 45.Vinther
BBrahe Pedersen
CElbrond
O Otitis media in childhood: sociomedical aspects with special reference to day care conditions.
Clin Otolaryngol. 1984;93- 8
Google ScholarCrossref 46.Blakley
BWBlakley
JE Smoking and middle ear disease: are they related? a review article.
Otolaryngol Head Neck Surg. 1995;112441- 446
Google ScholarCrossref 47.Gulya
AJ Environmental tobacco smoke and otitis media.
Otolaryngol Head Neck Surg. 1994;1116- 8
Google Scholar 48.DiFranza
JRLew
RA Morbidity and mortality in children associated with the use of tobacco products by other people.
Pediatrics. 1996;97560- 568
Google Scholar 49.Strachan
DPCook
DG Health effects of passive smoking, 4: parental smoking, middle ear disease, and adenotonsillectomy in children.
Thorax. 1998;5350- 56
Google ScholarCrossref 50.Stathis
SLO'Callaghan
MWilliams
GMNajman
JMAndersen
MJBor
W Maternal cigarette smoking during pregnancy is an independent predictor for symptoms of middle ear disease at 5 years' postdelivery.
Pediatrics. 1999;104E16
Google ScholarCrossref 51.Lewis
KWBosque
EM Deficient hypoxia awakening response in infants of smoking mothers.
J Pediatr. 1995;127691- 699
Google ScholarCrossref 52.Windham
GCHopkins
BFenster
LSwan
SH Prenatal active or passive tobacco smoke exposure and the risk of preterm delivery or low birth weight.
Epidemiology. 2000;11427- 433
Google ScholarCrossref 53.Wisborg
KKesmodel
UHenriksen
TBOlsen
SFSecher
NJ A prospective study of smoking during pregnancy and SIDS.
Arch Dis Child. 2000;83203- 206
Google ScholarCrossref 54.Dezateux
CStocks
JDundas
IFletcher
ME Impaired airway function and wheezing in infancy: the influence of maternal smoking and a genetic predisposition to asthma.
Am J Respir Crit Care Med. 1999;159403- 410
Google ScholarCrossref 55.Gergen
PJFowler
JAMaurer
KRDavis
WWOverpeck
MD The burden of environmental tobacco smoke exposure on the respiratory health of children 2 months through 5 years of age in the United States: Third National Health and Nutrition Examination Survey, 1988 to 1994.
Pediatrics. 1998;101E8
Google ScholarCrossref 56.Gilliland
FDBerhane
KMcConnell
R
et al. Maternal smoking during pregnancy, environmental tobacco smoke exposure, and childhood lung function.
Thorax. 2000;55271- 276
Google ScholarCrossref 57.Hanrahan
JPTager
IBSegal
MR
et al. The effect of maternal smoking during pregnancy on early infant lung function.
Am Rev Respir Dis. 1992;1451129- 1135
Google ScholarCrossref 58.Hoo
AFHenschen
MDezateux
CCosteloe
KStocks
J Respiratory function among preterm infants whose mothers smoked during pregnancy.
Am J Respir Crit Care Med. 1998;158700- 705
Google ScholarCrossref 59.Hu
FBPersky
VFlay
BRZelli
ACooksey
JRichardson
J Prevalence of asthma and wheezing in public schoolchildren: association with maternal smoking during pregnancy.
Ann Allergy Asthma Immunol. 1997;7980- 84
Google ScholarCrossref 60.Jedrychowski
WFlak
E Maternal smoking during pregnancy and postnatal exposure to environmental tobacco smoke as predisposition factors to acute respiratory infections.
Environ Health Perspect. 1997;105302- 306
Google ScholarCrossref 61.Lux
ALHenderson
AJPocock
SJ Wheeze associated with prenatal tobacco smoke exposure: a prospective, longitudinal study. ALSPAC Study Team.
Arch Dis Child. 2000;83307- 312
Google ScholarCrossref 62.Rusconi
FGalassi
CCorbo
GM
et al. Risk factors for early, persistent, and late-onset wheezing in young children. SIDRIA Collaborative Group.
Am J Respir Crit Care Med. 1999;1601617- 1622
Google ScholarCrossref 63.Stick
SMBurton
PRGurrin
LSly
PDLeSouef
PN Effects of maternal smoking during pregnancy and a family history of asthma on respiratory function in newborn infants.
Lancet. 1996;3481060- 1064
Google ScholarCrossref 64.Castellazzi
AMMaccario
RMoretta
A
et al. Effect of active and passive smoking during pregnancy on natural killer-cell activity in infants.
J Allergy Clin Immunol. 1999;103172- 173
Google ScholarCrossref 65.Mercelina-Roumans
PEBreukers
RBUbachs
JMvan Wersch
JW Hematological variables in cord blood of neonates of smoking and nonsmoking mothers.
J Clin Epidemiol. 1996;49449- 454
Google ScholarCrossref 66.Cederqvist
LLEddey
GAbdel-Latif
NLitwin
SD The effect of smoking during pregnancy on cord blood and maternal serum immunoglobulin levels.
Am J Obstet Gynecol. 1984;1481123- 1126
Google ScholarCrossref 67.US National Center for Health Statistics, Plan and operation of the Third National Health and Nutrition Examination Survey, 1988-1994.
Vital Health Stat 1. 1994;321- 417
Google Scholar 68.Gunter
EWLewis
BGKoncikowski
SM Laboratory Procedures Used for the Third National Health and Nutrition Examination Survey, 1988-1994. Atlanta, Ga Centers for Disease Control and Prevention, Center for Environmental Health1996;
69.Zhang
JYu
KF What's the relative risk? a method of correcting the odds ratio in cohort studies of common outcomes.
JAMA. 1998;2801690- 1691
Google ScholarCrossref 70.Feinstein
AR Multivariable Analysis: An Introduction. New Haven, Conn Yale University Press1996;
71.Shah
BVBarnwell
BGBieler
GS SUDAAN User's Manual, Release 7.5. Research Triangle Park, NC Research Triangle Institute1997;
72.Laupacis
ASackett
DLRoberts
RS An assessment of clinically useful measures of the consequences of treatment.
N Engl J Med. 1988;3181728- 1733
Google ScholarCrossref 73.Adams
PFMarano
MA Current estimates from the National Health Interview Survey, 1994.
Vital Health Stat 10. 1995;1931- 520
Google Scholar 74.Daly
KALindgren
BGiebink
GS Validity of parental report of a child's medical history in otitis media research.
Am J Epidemiol. 1994;1391116- 1121
Google Scholar 75.Pless
CEPless
IB How well they remember: the accuracy of parent reports.
Arch Pediatr Adolesc Med. 1995;149553- 558
Google ScholarCrossref 76.Alho
OP The validity of questionnaire reports of a history of acute otitis media.
Am J Epidemiol. 1990;1321164- 1170
Google Scholar 77.Caraballo
RSGiovino
GAPechacek
TFMowery
PD Factors associated with discrepancies between self-reports on cigarette smoking and measured serum cotinine levels among persons aged 17 years or older: Third National Health and Nutrition Examination Survey, 1988-1994.
Am J Epidemiol. 2001;153807- 814
Google ScholarCrossref