Chapter 6

Respiratory Health Effects

6.0 Introduction

The relationship between ETS exposure and a variety of nonmalignant respiratory tract health endpoints has been examined extensively in the epidemiologic and experimental literature. Among children, the most common outcomes studied include asthma induction and exacerbation, alterations in lung development, and otitis media and chronic middle ear effusions in children. Among adults, endpoints of interest have included lower respiratory tract symptoms, lung function, and acute irritative symptoms of the upper respiratory tract. Each of the lower respiratory tract endpoints, as well as otitis media, were reviewed in reports by the Surgeon General's Office (U.S. DHHS, 1986), the National Research Council (NRC, 1986), and most recently by the U.S. Environmental Protection Agency (U.S. EPA, 1992); upper respiratory tract irritation and sensory annoyance were reviewed in the Surgeon General's and NRC reports only. This chapter synthesizes the data considered in these prior literature reviews with results from more recent studies in order to assess the possible relationship between ETS exposure and each of the above-mentioned health endpoints.

Asthma is a chronic respiratory condition characterized by airway inflammation and episodic airflow limitation. Depending on the clinical definition used, about 2-3% of adults and up to 10% of children may be affected (Evans et al., 1987; Schwartz et al., 1990; Gergen et al., 1990; Gerstman et al., 1989). Asthma is the most common chronic condition of childhood, and in 1990 accounted for approximately $6.2 billion in health care expenditures nationally (CDC, 1992). No similar cost estimate is available for California; however, in the state in 1993 there were approximately 43,000 admissions to hospital with a primary diagnosis of asthma, about 18,600 of which were for children under age 18 (Personal communication: Dr. Marvin Bohnstedt, California Department of Health Services). Typical symptoms of asthma include cough, chest tightness, difficulty breathing, and wheezing. One of the hallmarks of asthma is airway hyperresponsiveness, an exaggerated tendency of the airways to constrict in response to physical stimuli, such as cold, dry air, or chemical agents, such as methacholine or histamine.

In its recent review, the U.S. EPA (1992) concluded that, "There is now sufficient evidence to conclude that passive smoking is causally associated with additional episodes and increased severity of asthma in children who already have the disease" (p. 248). This conclusion appears to have been based on a review of several studies summarized in Table 6.1. These and additional relevant studies are described in the following pages.

6.1.1.1 Epidemiologic evidence

Evans and colleagues (1987) analyzed data on 276 children from low-income families receiving care for asthma at outpatient clinics of several New York City hospitals. Information on ETS exposure was obtained during interviews of the parent or guardian and the child. Data on emergency room (ER) visits and hospitalizations were obtained by reviewing records for a one-year period, data on lung function were obtained during a random clinic visit occurring within one year after the interviews. Eight children who admitted to active smoking were excluded from the analysis, as were 77 other children with incomplete data. The relationships between the child’s ETS exposure (as reported by the parent or guardian) and the outcome variables were analyzed by multiple regression techniques in which the influence of up to 34 potential confounders and effect modifiers were considered, including age, gender, ethnicity, several indicators of socio-economic status, indices of medical management, and an index of residential allergen and irritant exposure, among others. Evans and co-workers reported that ER visits were positively associated with reported ETS exposure (p<0.01). The mean annual frequency of ER visits among the children exposed to ETS in the home was 3.09 ± 0.40, while that for children not exposed was 1.83 ± 0.29. Thirty-nine percent of children from nonsmoking homes made no ER visits, compared with 29% of children from households with smokers. In addition, 13% of children from nonsmoking homes made four or more ER visits, compared with 32% of children from smoking households.

Evans et al. also reported, however, that ETS exposure was not associated with either hospitalizations or with percent predicted lung function. While ostensibly inconsistent with the ER results, there were relatively few hospitalizations during the period of observation (191 children x 0.20 (mean) hospitalizations/child/yr » 38), resulting in a low power to detect an effect. Furthermore, to the extent that nondifferential misclassification of exposure occurred (or that some smokers may have reported that they were nonsmokers), this could bias the analysis against finding an effect. (Either of these types of misclassification of exposure would also tend to diminish the reported relationship between ETS exposure and ER visits.) The report does not provide enough information to evaluate the lack of association of ETS exposure and lung function. The protocol for pulmonary function test (PFT) administration is not well described; for example, it is not clear whether any children were experiencing an asthma flare when tested. Here also misclassification of exposure status could also create a bias against finding an association.

More recently, Chilmonczyk and colleagues (1993) undertook a study of similar design to that of Evans et al.(1987), using more sensitive indicators of asthma exacerbations (review of medical records at a large allergy/asthma clinic) and of ETS exposure (measurement of urinary cotinine at enrollment in addition to parental questionnaire). Cotinine is the major metabolite of nicotine, and is a good integrated indicator of recent ETS exposure (1-2 days; See chapter on Exposure Measurements and Prevalence). Review of medical records was done by observers blinded as to the children’s ETS exposure status. Of the 199 children (aged 8 mos. to 13 yr, mean » 7.5 yr) enrolled in the study, 145 were old enough to undergo pulmonary function testing. Whether assessed by urinary cotinine or by parental reporting, ETS exposure was found to be associated with increased frequency of asthma exacerbations in a dose-dependent manner. Using multiple regression techniques that adjusted for the child’s age, gender, day-care attendance, the mother’s age and educational level, the investigators reported relative risks for the highest versus the lowest exposure categories of 1.7 (95% CI = 1.4-2.1) for exposure assessed by cotinine and 1.8 (95% CI = 1.4-2.2) for parent-reported exposure. Pulmonary function tests reported as percent predicted FEV₁ (forced expiratory volume in one second -- a measure of lung volume and central airway caliber) and FEF_25-75 (expiratory flow during the middle half of a forced vital capacity (FVC) maneuver -- an indicator of the caliber of the more peripheral, mid-sized to smaller airways) were decreased with increased ETS exposure in a dose-dependent manner, with urinary cotinine as the exposure indicator.

In this investigation, parental reports of no ETS exposure were consistent with the cotinine results 86% of the time, while the concordance of reported exposure and cotinine measurements was 77%. Henderson et al. (1989) have previously shown that cotinine levels in preschool children tend to be stable over at least a four-week period, presumably due to regular daily patterns of ETS exposure. Ogborn et al. (1994) reported similar findings in children aged 3 to 11 (see below). To the extent that the single urinary cotinine measurement in the study by Chilmonczyk and colleagues was an accurate reflection of longer-term exposure, this study suggests an exposure-related chronic effect on exacerbations of pediatric asthma and on indices of lung function.

Murray and Morrison (1989) examined 419 children aged 1 to 17 attending an allergy clinic in Vancouver, British Columbia. At the initial visit a trained interviewer administered a standardized questionnaire to the parent and the patient containing questions regarding the child’s asthma history, symptoms and medication use, other respiratory illnesses, and a variety of potential residential exposures (ETS from one or both parents, whether a woodstove was used for home heating or a gas stove for cooking, the presence of cats or dogs). The children were asked privately if they were themselves active smokers: the four that admitted to this were excluded from the analysis. The investigators created an asthma severity score based on questionnaire responses. In addition, the patients had allergy testing (by skin prick), spirometry (for those ³ 6 year old), and an examination of bronchial reactivity to histamine (in children ³ 7 year old). Children of smoking mothers (n=92) had more severe asthma than children of nonsmoking mothers (n=322), as evidenced by an increased asthma score, greater airway reactivity, as well as decreased FEV₁ and FEF_25-75 (for all these differences, p<0.01). These results were driven by the effects observed in boys; for girls, only the asthma severity score was significantly increased with smoking versus nonsmoking mothers.

Murray and Morrison found that differences between asthmatic children of smoking versus nonsmoking mothers became more pronounced with increasing age (or duration of exposure). In multivariate regressions of these indices of severity on several independent variables (recent respiratory infection, recent bronchodilator use, positive skin prick test, presence of gas stove or wood stove, and the number of cigarettes smoked at home by either parent), stratified on age and/or gender, maternal smoking repeatedly emerged as one of the two strongest predictors of asthma severity. Paternal smoking was generally without effect.

The results of this study were consistent with earlier publications by the same investigators using subsets of this study population (Murray and Morrison, 1986 and 1988). In an analysis of 240 subjects for whom data were collected as described above (for the 1989 publication), Murray and Morrison found a highly significant association between indices of increased asthma severity (here limited to airway reactivity, FEV₁, and FEF_25-75) and maternal smoking, with no such associations observed for paternal smoking. Hypothesizing that the children’s asthma would be exacerbated during the colder, wetter months (when homes would be kept more tightly closed and children would spend more time indoors, thereby increasing the intensity of ETS exposure), they stratified the analysis by season (October through May versus June through September). They observed no differences in any of the above indices between children of smoking and nonsmoking mothers during the dry season, but found highly significant differences in FEV₁, FEV_25-75, and airway reactivity during the wetter months. Moreover, during the wet, but not the dry, season there was evidence of an exposure-response relationship between the number of cigarettes smoked by the mother in the home with each of these indices. These relationships were corroborated by multiple regressions of FEV₁, FEF_25-75, and airway reactivity on age, duration of asthma, gender, recent respiratory infection, recent medication use, positive skin allergy test, family history of asthma, presence of pets, heating type, presence of a gas range, number of siblings, and parental smoking.

More recently, Murray and Morrison (1993) expanded their analysis to include 807 nonsmoking asthmatic children and adolescents meeting the same eligibility requirements and for whom similar data were collected as in these authors’ prior reports. In this analysis, they compared indices of asthma severity in children first attending the asthma clinic before July 1986 with those attending afterward. They reported that, among children with at least one smoking parent, reported daily exposures to cigarettes smoked in the same room were markedly lower after 1986 (3.4 after versus 6.6 before for smoking mothers (p=0.005) and 2.0 after versus 4.6 before for fathers (p=0.001)). Concomitant with this apparent decline in exposure, children of smoking parents entering the study after July 1986 had asthma that was less severe than those entering earlier, as manifested by significant improvements in the asthma score, FEV₁ and FEF_25-75. Among children of nonsmoking parents who enrolled after July 1986, the asthma severity score was not significantly different, but both FEV₁ and FEF_25-75 were increased, though less so than among the children of smoking parents. Improvements in pulmonary function test values occurred regardless of the smoking status of either parent; however, a significant difference in the asthma severity score was observed only among the children of smoking mothers. Modeling the spirometric indices as a function of numbers of cigarettes smoked in the same rooms as the child(ren), controlling for several relevant covariates (age, gender, age of asthma onset), Murray and Morrison found that the differences among smokers’ children across the time periods decreased, which would be expected if exposure to parents’ smoking is one of the etiologic factors underlying the difference. Airway reactivity showed no marked differences in before/after comparisons within either group of children. In light of the significant improvement in the other indicators of asthma severity, the lack of a significant change in airway reactivity was unexpected; the authors had no ready explanation for this anomaly other than that the persistent hyperresponsiveness of these asthmatics’ airways might not have been affected by recent decreases in exposure to parental ETS.

O’Connor et al. (1987) examined the relationship between parental smoking and airway reactivity in 286 children, aged 6 to 21 yr (mean » 13 yr), in East Boston. Airway responsiveness was assessed by measuring FEV₁ before and after challenge with subfreezing air. Of the 21 asthmatic study subjects, those with smoking mothers showed a greater reduction in FEV₁ in response to cold air challenge than did those with nonsmoking mothers (24 ± 3.3% vs 11.9 ± 4.8%, respectively; p = 0.07). In multiple linear regression models examining cold air-induced change in FEV₁ as the dependent variable in relation to nine putative independent variables, maternal smoking emerged as a significant (p = 0.02) predictor of D FEV₁, after adjusting for predicted FEV₁. In a stepwise multiple linear regression, maternal smoking and predicted FEV₁ were the only two variables to enter the model. Paternal smoking was unrelated to bronchial responsiveness in any analysis. The small sample size of the asthmatic children and adolescents in this study limits both the statistical significance and the generalizability of the findings. Nevertheless, the trend observed in this investigation is consistent with the results reported by Murray and Morrison (1989).

Household ETS exposure may affect severity of asthma in adults as well as children. Jindal (1994) investigated several measures of respiratory morbidity in 200 never-smoking adult asthmatics, aged 15 to 50, attending an outpatient chest clinic in India. Information on ETS exposure and on various indices of asthma control during the preceding year were obtained by questionnaire during a clinic visit. ETS-exposed participants (n = 100) were defined as those who reported a minimum of one hour exposure/day, or seven hour/week, for at least one year. Indices of asthma control included: lung function measurements, the use and number of maintenance bronchodilator medications, requirement for corticosteroids (presumably orally administered, though this is not clear from the article), numbers of visits to the emergency department, admissions to hospital, number of acute episodes, and number of times that the patients used parenterally administered (i.e., via injection) asthma medications at home (reportedly a common practice in the study area). Pulmonary function testing was done within 24 hours of the clinic visit at which the questionnaire was administered. In comparison with a nonexposed group of 100 patients, the ETS-exposed group showed significantly lower forced expiratory lung function indices (FEV₁, FEV₁/FVC, and FEF_25-75%). In addition, though the numbers of patients in the two groups did not differ with respect to most of the morbidity measures, the ETS-exposed group included significantly more patients on maintenance bronchodilator therapy and corticosteroids required to control symptoms. When expressed on a per-patient-per-year basis, however, all the indices in the ETS-exposed group were significantly higher, except for the numbers of hospital admissions and weeks of bronchodilator use per patient. Although not conducted or analyzed as meticulously as the investigation by Chilmonczyk and colleagues (1993) (above), this report suggests that regular ETS exposure may affect control and severity of asthma in adults as well as children.

The above reports support the existence of an association of chronic or repeated ETS exposure with severity of asthma measured by a variety of indices. In several epidemiologic studies, ETS has been implicated as a risk factor for exacerbation of asthma, measured as increases in symptoms, medication use, and clinic or emergency room (Evans et al., 1987; Chilmoncyzk et al., 1993; Jindal et al., 1994; Ostro et al., 1994 (see below)). Airway responsiveness, one indicator of asthma severity, tends to be increased in asthmatic children whose mothers smoked in comparison with those with nonsmoking mothers (O'Connor et al., 1987; Murray and Morrison, 1989). The results of controlled chamber investigations suggest that even single exposures of adult asthmatics to ETS can elicit prolonged airway hyperresponsiveness (AHR), which provides experimental support for the epidemiological observations (Menon et al., 1992). Increased airway responsiveness facilitates bronchoconstriction (and the concomitant symptoms of chest tightness, wheeze, and difficulty breathing) in response to respiratory irritants, such as ETS (NRC, 1986). The above findings support the assessment articulated by the U.S. EPA that there is sufficient evidence to support the inference of a causal relationship between ETS exposure and "additional episodes and increased severity of asthma in children who already have the disease."

Whether acute ETS exposure can precipitate a specific asthma flare is not so clear-cut, however. Ehrlich et al. (1992) undertook a case-control study of 72 children visiting the emergency room (ER) for their asthma, 35 children attending an asthma clinic who were not acutely ill, and 121 nonasthmatic control children. ETS exposure was assessed by questionnaire and by urinary cotinine/creatinine ratios (CCR). Using a cut-point of 30 ng cotinine/mg creatinine to distinguish exposed versus unexposed children, they found no difference in recent ETS exposure between asthmatics recruited from the ER and those from the clinic. In contrast, the ETS exposure odds ratio for all asthmatics versus controls was 1.9 (95% CI = 1.04 - 3.35). The mean CCR in the acute asthmatics (46.2 ± 98.3) was greater than that in the nonacute asthmatics (38.5 ± 74.1), but this difference was not statistically significant. From the questionnaire responses, there was no significant difference between acute and nonacute asthmatics in relation to maternal smoking, whereas the exposure odds ratio for asthmatics versus controls was 2.0 (95% CI = 1.1 - 3.4). This investigation suggests that ETS exposure is a risk factor for clinical asthma but, in this study population, may not have been a significant precipitant of asthma flares serious enough to warrant a visit to the ER. However, the investigation by Ehrlich et al. cannot adequately address the latter issue because of limited statistical power (<50% probability of detecting a 2-fold exposure difference between acute and nonacute asthma), coupled with the likelihood that the children recruited from the clinic had more severe asthma (with 80% on daily asthma medication vs. 36% of the ER patients).

Ogborn and colleagues (1994) also investigated whether there was an association between exacerbations of asthma and acute exposure to ETS. Data consisted of parental responses to detailed ETS exposure questionnaires and measurements of urinary cotinine obtained from children, aged 3 to 11, who were seen during a visit to the ER or primary-care clinic during an acute asthma flare and at a follow-up clinic visit after the flare had subsided. The investigators found no significant difference between mean urinary cotinine values (± standard deviation) at the acute versus the well visit (81 ± 62 ng/ml and 77 ± 57 m g/ml, respectively). Similarly, the mean CCRs at the acute versus the well visits were 93 ± 109 ng/ml and 97 ± 87 ng/ml, respectively. (Note that the mean CCR values in this study are at least twice as great as those reported in the Ehrlich et al. (1992) study, suggesting heavy ETS exposure.) In this population, the prevalence of household smoking was remarkably high (77% overall, and 63% among the children’s mothers). However, this investigation is also limited by low power: the sample size of 56 had a power of 0.80 (a = 0.05) to detect the change in CCR expected to result from a 20-cigarette per day change in ETS exposure. In other words, this study would have the power to detect a difference equivalent to household smokers’ (or other sources of ETS exposure) reducing their consumption of cigarettes by one pack a day, over a several-week interval, with no concerted smoking cessation intervention. Thus, this underpowered study cannot address the issue of whether acute ETS exposure can provoke an exacerbation of asthma.

However, the study by Ogborn et al. does provide additional interesting information about the stability of CCRs over time. The acute and well visits of these children were separated by approximately three to four weeks (detailed information on the timing of the visits is not provided in the report). Among the children reported to be "exposed" to ETS, the mean CCR was 105 ± 119 ng/ml at the ER visit and 105 ± 85 ng/ml at the follow-up visit, which is consistent with regular ongoing exposures. In addition, this study provides evidence that parents may under-report ETS exposure. Using a urinary CCR of 30 ng/mg as a cut-point for recent ETS exposure, the mean CCR levels in the children reported not to have been exposed at the acute and well visits were 41 ± 30 ng/ml and 83 ± 100, respectively.

There is suggestive recent evidence that ETS exposure may elicit acute symptoms in adults. Ostro et al. (1994) investigated the relationships between exposures to indoor combustion products and daily symptoms in a population of adult asthmatics residing in Denver, Colorado. This study included 164 subjects, many of whom had moderate to severe asthma, and some of whom experienced respiratory infections and asthma flares during the period of observation. This investigation also had more than 10,000 observations, which afforded substantial statistical power to detect associations with indoor exposures, including ETS. Both symptom and exposure data were recorded by the study participants in an intake questionnaire and in daily diaries over a three-month period. In multiple logistic regression models corrected for serial correlation and repeated measures, these investigators reported an odds ratios of 1.61 (95% CI = 1.06 - 2.46) for restricted activity days in relation to ETS exposure. They also reported significantly increased odds ratios for the occurrence of moderate to severe cough and shortness of breath, which were still elevated but no longer significant after correction for autocorrelation. However, having a smoker in the home during the course of the study corresponded to an odds ratio of 2.05 (95% CI = 1.78 - 2.40) for increased daily moderate to severe shortness of breath, suggesting a relationship of chronic exposure to acute symptoms as well.

The studies reviewed in this section support the previous finding by the U.S. EPA (1992) that there is "sufficient evidence¼ that passive smoking is causally associated with additional episodes and increased severity of asthma in children who already have the disease." There is suggestive evidence that ETS exposure may exacerbate adult asthma. The U.S. EPA (1992) estimated that ETS exposure potentially could exacerbate pre-existing asthma in approximately 20% of 2 to 5 million children, i.e., in 0.4 to 1 million children. Assuming that 12% of those children reside in California would result in estimates of 48,000 to 120,000 asthmatic children who could experience a worsening of their condition due to exposure to ETS.

6.1.1.2 Evidence from chamber studies

Several chamber studies have investigated potential relationships between controlled exposure to ETS and lung function and airway reactivity in asthmatic subjects. The results of these investigations are summarized in Table 6.2. Experimental exposure of human volunteers to various pollutants under controlled laboratory conditions can provide useful pathophysiological information. The principal advantages of this methodology over epidemiological studies is that exposure to the pollutant(s) of interest can, in theory, be precisely measured, and thus exposure-response relationships determined. While exposure conditions can also be controlled in animal experiments, the obvious strength of human chamber studies is that no cross-species extrapolation is required. On the other hand, microscopic or biochemical examination of pollutant-induced tissue damage is more limited in humans by both ethical and practical considerations. However, controlled human exposures are also subject to the following structural limitations: (1) only short-term responses to relatively brief exposures (i.e., minutes to hours) can be evaluated; (2) there is often limited statistical power to detect effects, due to the typically small number of subjects; (3) controlling the experimental conditions may result in failure to capture effects found in complex real-world exposures; (4) multiple selection biases in recruiting volunteers reduce the generalizability of such studies (e.g., systematic exclusion of people with a history of recent respiratory infection; relatively few studies of children, adolescents or other potentially susceptible subgroups). It should be emphasized, however, that these limitations all tend to underestimate pollutant effects. Given the potential shortcomings of such investigations, negative findings may in some cases reflect the constraints of study design more than biological reality.

In controlled exposure studies, volunteer subjects are exposed to one or more pollutants through a mouthpiece (oral breathing only) or in a chamber (oronasal breathing). ETS-related studies of asthmatics have mainly been conducted in exposure chambers with resting subjects. Data collected usually have included graded respiratory symptoms and a variety of indices of pulmonary function, such as the amount of air a subject can exhale in one second after a deep inspiration (FEV₁) or the lung's resistance to airflow (airway resistance (R_aw) or specific airway resistance (SR_aw)). Several studies of asthmatics involving ETS exposures have also examined airway responsiveness (also known as bronchial reactivity) (described below).

Chronic airway inflammation and episodic, reversible bronchoconstriction are hallmarks of asthma. Inflammation is associated with bronchial hyperreactivity or hyperresponsiveness, which refers to an exaggerated tendency of the airways to constrict when exposed to respiratory irritants or other substances. Airway hyperresponsiveness (AHR) is also observed in many persons with emphysema and bronchitis and in otherwise healthy individuals during and after respiratory tract infections and after exposure to respiratory irritants such as ozone. In general, however, such reactivity is markedly greater in asthmatics compared with nonasthmatics. Airway responsiveness to numerous stimuli can be measured in clinical studies. Methods used to induce and measure nonspecific bronchial reactivity in asthmatics include exercise or hyperventilation with cold or dry air, or inhalation of pharmacological agents (e.g., histamine or methacholine). Although these pharmacological agents also cause bronchoconstriction in healthy individuals, asthmatic airways constrict at much lower exposure concentrations. AHR creates the potential for a flare or exacerbation of asthma, with heightened bronchial responses to other nonspecific airborne irritants.

The series of studies conducted at Tulane University (Stankus et al. 1988; Menon et al., 1991 and 1992) suggest that a substantial fraction of asthmatics with self-reported sensitivity to ETS also appear to demonstrate susceptibility by more objective means of assessment (tests of lung function and airway responsiveness). While such susceptible individuals have increased baseline AHR (as measured by methacholine challenge testing), nonspecific airway reactivity does not fully explain this sensitivity, since other asthmatics with increased AHR do not show marked reactions to ETS inhalation challenge. Although the physiologic basis for susceptibility is not well understood, the effects of exposure on such ETS-reactive and nonreactive individuals appear to be reproducible, suggesting the existence of intrinsic individual characteristics (Stankus et al., 1988; Menon et al., 1991). Though these acute studies cannot replicate exposure conditions

experienced by free-living subjects, the findings of increased AHR support the epidemiological studies described earlier, which indicate that repeated household ETS exposures tend to result in worse control of asthma.

Most of the ETS inhalation chamber studies show slight-to-moderate, transient effects on lung function in at least some of the study subjects. In several studies, participants experienced decrements in lung function exceeding 20%, which would be considered clinically significant, particularly in conjunction with the occurrence of lower respiratory symptoms such as chest tightness, dyspnea, and cough. To the extent that subjective symptoms of asthma were reported, clinically meaningful respiratory symptoms were identified in some participants in several studies (Knight and Breslin, 1985; Dahms, 1981, Stankus et al., 1988, Shephard et al., 1979); however, this was clearly not a universal finding (Magnussen et al., 1993, Oldigs et al., 1991). AHR occurring after ETS exposure was also reported inconsistently in these studies; nevertheless, the only studies that examined delayed AHR at times that would be likely to detect the effects of an inflammatory response did find significant ETS-associated increases (Menon et al., 1991 and 1992).

The controlled exposure studies do not clearly demonstrate a consistent effect of acute ETS exposure on asthmatics as a whole. As noted above, however, general design constraints in such studies militate against finding effects, e.g., small sample size, systematic exclusion of potential participants who have recently been ill or those with brittle asthma, acute exposures only. Each of these studies has one or more weaknesses in design or analysis; thus, neither individually nor collectively can these investigations definitively address the issue of whether acute ETS exposure can precipitate an asthma flare. For instance, in Shephard et al. (1979), anti-asthma medications were not withheld prior to the exposures in 13/14 participants, which would be likely to dampen any potential effects of ETS. Several investigations involved fewer than a dozen subjects (Knight and Breslin, 1985, Wiedemann, 1986; Oldigs et al., 1991); all but two of the remaining studies had fewer than two dozen subjects. In at least two studies (Oldigs et al., 1991; Magnussen et al., 1993), the participants were also regularly exposed to ETS at home, which could affect their responses in an acute experimental setting. Menon et al. (1991) indicated that at least one hour of exposure was needed to elicit respiratory responses, even among their ETS-sensitive subjects. Several of the chamber studies (including most of the "negative" ones) involved exposure to ETS that was limited to only one hour (Knight and Breslin, 1985, Wiedemann, 1986; Oldigs et al., 1991, Dahms, 1981, Magnussen et al., 1993, Danuser et al., 1993). Moreover, there is considerable variability among asthmatics with respect to susceptibility to airborne irritants, including ETS. For instance, adult asthmatics vary at least seven-fold in their susceptibility to the bronchoconstrictive effects of sulfur dioxide, which is probably the most well-studied respiratory irritant in relation to asthma (Horstman et al., 1986). Thus, even apart from differences in study design and experimental conditions, investigations of the effects of acute ETS exposure in asthmatics would be expected to produce variable results.

Finally, one criticism of ETS chamber studies has been that the characteristic odor and mucous membrane irritation make it difficult to blind the participants to the nature of the exposure (ETS vs. clean air). This in turn is hypothesized to result in psychological suggestion as a cause of observed symptoms, changes in lung function, and so forth (Witorsch, 1992). Similarly, it has been argued that the physical conditions of participating in an inhalation challenge study create "stress" to which any positive results might be attributed (Witorsch, 1992). As for the latter observation, the use of control exposures, control subjects or both are intended to provide a basis for "control" for whatever stresses are associated with the experimental procedure. Suggestibility related to the lack of blinding may theoretically augment symptomatic and physiological responses, but experimental evidence suggests that, if present, its influence is weak. In the study by Danuser et al. (1993), the subjects wore noseclips and had the ETS administered by mouthpiece, essentially blinding them to the differences in concentration of ETS delivered. Yet most of the symptomatic responses of the subjects with AHR, though not clinically severe, appeared to be dose-related, which would be difficult to attribute to suggestion. Urch et al. (1988) investigated the role of suggestibility in 40 nonsmokers, including 16 asthmatic and 24 nonasthmatic adults. Participants were exposed in an exposure chamber on separate occasions for 65 minutes to clean air ("sham"), moderate or heavy smoke (17 and 31 ppm carbon monoxide, respectively). Though they viewed a bank of burning cigarettes outside the chamber on all occasions, the smoke was diverted during the sham experiment. These investigators reported, among other results, the occurrence of significant dose-related symptoms in asthmatics and nonasthmatics and a dose-response relationship for several measures of lung function. They also administered a battery of psychometric tests to assess the subjects’ suggestibility, and found little correlation between physiological changes and indices of suggestibility. Assuming the subjects were unable to distinguish between the moderate and heavy smoke concentrations, Urch et al. concluded that the dose-response relationships were more likely of physiological than psychological origin, although the latter may have played a minor role in the observed responses.

In summary, although the design constraints of the chamber studies limit the interpretation of the results, they do suggest that there is likely to be a subpopulation of asthmatics who are especially susceptible to ETS exposure. The physiological responses observed in these investigations appear to be reproducible in both "reactors" and "nonreactors". It is unlikely that the physiological and symptomatic responses reported are due exclusively to either stress or suggestion.

Infections of the respiratory tract are the most common acute illness of childhood. Apart from the morbidity (and occasional mortality) attributable to respiratory infections, they also represent risk factors for asthma (both induction and exacerbation of existing disease) and possibly other chronic respiratory effects in later life (Burrows et al., 1977; Gold et al., 1989; Henderson et al., 1992; Schroekenstein and Busse, 1988). The relationship of parental smoking to the risk of respiratory infection has been extensively investigated. It has been clearly established in nearly two dozen reports reviewed by the NRC (1986), the Surgeon General (1986) and the U.S. EPA (1992) that ETS exposure increases the risk of acute lower respiratory disease in young children by 1.5 to 2-fold.

The estimates of the magnitude of the effect of household ETS exposure on respiratory infections are remarkably consistent among the many studies that have examined this relationship. The effects are most marked in infants and toddlers, and are often not detectable in school children, who may be less exposed than younger children or who may have developed immunity against many respiratory pathogens. Several studies noted the existence of a dose-response relationship, where dose was measured by the number of cigarettes smoked in the household. In studies conducted mainly in Europe and North America, maternal smoking has repeatedly been found to bear a stronger relationship to respiratory illness than paternal smoking. This is likely to be due to the greater amount of time spent by mothers than by fathers with young children, enhancing the frequency and intensity of ETS exposure, strengthening the inference of a causal association. However, the series of reports by Chen et al. (1986, 1988, 1989), which involved cohorts of infants with nonsmoking mothers, also found dose-response relationships with paternal smoking. The reviews by the U.S. EPA, NRC, and the Surgeon General all noted that most of these studies, while not free of all sources of bias, had adjusted for many identifiable confounding variables and found that the ETS effects were independent of sex, race, maternal age, socio-economic status (SES), residential crowding, and number of siblings. In some studies, breast-feeding had a protective effect, as did day-care attendance, the latter presumably by decreasing exposure to parental ETS. Low birth weight increased susceptibility to ETS effects (U.S. EPA 1992). As a group, these nearly two dozen investigations are quite consistent and provide convincing evidence of an increased risk of lower respiratory illness in young children.

The discussion of the relationship between ETS exposure and pediatric respiratory illness has been adequately addressed in the reviews by the NRC, the Surgeon General, and the U.S. EPA., and therefore a de novo analysis of the primary literature has not been undertaken. More recent published investigations support the conclusions articulated in these reviews. For example, Chen (1994) reported increased risks of hospitalization for respiratory illness during the first 18 months of life in China as follows: ORs = 2.91 (95% CI = 1.41 - 6.01) and 4.48 (95% CI = 2.07 - 9.73) among low birth weight infants exposed to light and heavy household smoking, respectively, and 1.40 (95% CI = 0.96 - 2.03) and 1.61 (95% CI = 1.08 - 2.41) for similar exposures among children of normal birth weight. Similarly, Robertson (1994) found an ETS-associated increased risk of hospitalization ( for respiratory and other causes) during the first six to ten months of life in a cohort of 1,877 infants in New Zealand (OR = 1.52, 95% CI = 1.08 - 2.14). In a prospective study of respiratory illness during the first two years of life in 836 Australian children, Douglas et al. (1994) reported that maternal smoking was associated with a significantly increased frequency of respiratory illness in the second, but not the first year of life. While each of these investigations has one or more methodological limitations, they are generally consistent with the reports discussed in the above-noted reviews. These and other recent studies support the conclusions stated in the reports by the NRC, the Surgeon General, and the U.S. EPA., that ETS exposure clearly confers an increased risk of acute lower respiratory disease in young children.

As noted above, ETS exposure in early childhood has been estimated to increase the risk of lower respiratory infection by 1.5 to 2-fold. On a national level, this magnitude of increased risk would correspond to 150,000 to 200,000 ETS-related cases of lower respiratory illness annually in children under 18 months of age (U.S. EPA, 1992). Noting that approximately 5% of patients with lower respiratory illness require hospitalization, the U.S. EPA estimated that 7,500 to 15,000 admissions to hospitals are attributable to ETS exposure each year in the U.S. These may be underestimates of effect, since the calculations on which they were based did not account for either exposure to paternal smoking or the likelihood of occurrence of repeated episodes of illness in regularly exposed children. If 12% of the population at risk resides in California, these estimates would correspond to 18,000 to 36,000 new cases of lower respiratory illness each year and 900 to 1,800 hospitalizations attributable to ETS exposure.

A number of studies, cited in the preceding subsections, link passive smoking with lower respiratory tract conditions in children. The relationship between ETS exposure and childhood upper respiratory tract conditions, particularly acute and chronic otitis media, constitutes a separate area of concern. This topic has been reviewed extensively by the Surgeon General's Office (U.S. DHHS, 1986), the NRC (1986), and the U.S. EPA (1992). This section briefly summarizes the findings of the above three reports, reviews those studies not included in the reports, and explores related evidence on pathophysiology.

6.1.3.1 Background/Definitions

Otitis media is the most commonly diagnosed problem in outpatient pediatrics in the United States today (Etzel et al., 1992). In the context of this discussion, it is useful to consider the anatomy and physiology of middle ear disease before reviewing the data concerning ETS as a risk factor for otitis media. The middle ear communicates with the nasopharynx via the Eustachian tube. The Eustachian tube acts as a barrier to microorganisms originating in the pharynx, as a pressure equalization channel, and as conduit of drainage for secretions originating in the middle ear. Eustachian tube dysfunction of whatever etiology results in a sustained pressure differential between the middle ear and the surrounding atmosphere, with subsequent effusion of serous fluid into the middle ear. Alone, this condition is called "serous otitis media," and produces a sensation of fullness and temporarily decreased hearing. Should the serous fluid become infected (usually with bacteria), "acute otitis media" results, with pain, fever, and the potential for tympanic membrane (TM) perforation. Serious secondary complications (meningitis, mastoiditis) can also occur, as can a self-perpetuating cycle of acute and serous otitis media (Goycoolea, 1991). Chronic serous effusions, with or without intervening infections, may lead to a variety of complications, including mucoid effusion (so-called "glue ear") and stretching of the tympanic membrane ("incompetent TM" or "atelectatic TM"), each resulting in more sustained hearing loss than does simple serous otitis. Tympanic membrane perforation can result, not only in hearing loss, but also in the formation of a "cholesteatoma" -- an ingrowth of squamous cells from the exterior of the TM -- which, in turn, can expand and destroy the ossicles of the middle ear. Hearing loss, whether from sustained serous otitis media, mucoid effusion, atelectatic TM, TM perforation, or ossicle destruction due to cholesteatoma, can result in communication difficulties and educational impairment in children.

6.1.3.2 Epidemiologic Data

The Surgeon General (1986) and NRC (1986) reviewed five and the U.S. EPA (1992) an additional ten studies on ETS exposure in childhood and upper respiratory tract conditions. Twelve of these fifteen studies examined acute or chronic otitis media and/or middle ear effusions. These twelve studies included five case-control, four prospective, and three retrospective or cross-sectional investigations, and are summarized in Table 6.3, as well as in the above three reviews. In all but three of these 12 studies, statistically significant relationships between exposure and outcome were apparent.

The reports of both the Surgeon General and the U.S. EPA expressed concern regarding potential misclassification of exposures based solely upon historical measures. Two studies (Strachan et al., 1989; Etzel et al., 1992) used objective measures of ETS exposure (salivary and serum cotinines, respectively), and both found a statistically significant relationship between ETS exposure and outcome. Likewise, two studies (Iverson, 1985; Etzel et al., 1992) employed periodic prospective screening for middle ear disease, thus eliminating differential utilization of medical services by parents as a possible confounder. Again, both of these studies found statistically significant associations between ETS exposure and middle ear disease.

Table 6.4 summarizes an additional ten epidemiologic studies not included in the above summary reports (the Surgeon General's Office, NRC, or U.S. EPA). Several of these additional studies were problematic with respect to their study designs. For example in the Kallail (1987) study, cases and controls were not subjected to the same screening procedures. In the Pönka (1991) study, parental reporting, rather than medical records or objective surveillance, was the index of disease outcome. Nevertheless, three of the studies reported nonsignificant positive associations between ETS and middle ear disease, five reported a significant relationship, and none reported a protective effect. None of these studies utilized biomarkers of ETS exposure. The one study that used an objective measure applied on a prospective basis (tympanometry) did report a slight, but nonsignificant association (Zielhuis, 1989).

6.1.3.3 Summary of Epidemiologic Data

The Surgeon General's report (U.S. DHHS, 1986) summarized the studies as "show[ing] an excess of chronic middle ear effusions and diseases in children exposed to parental smoke." The U.S. EPA (1992) report concluded that there was "good evidence demonstrating a significant increase in the prevalence of middle ear effusion in children exposed to ETS," but only "some evidence [for] acute middle ear infections" (acute otitis media). While the ten studies in Table 6.4 are collectively somewhat less supportive of an association between ETS exposure and middle ear disease than those previously reviewed by the Surgeon General's Office and the U.S. EPA, the study design of some of these studies was problematic.

Ten of the twelve studies reviewed by the Surgeon General's Office or the U.S. EPA reported statistically significant relationships between ETS exposure in the home and middle ear conditions in children. Of the additional ten studies reviewed here, five showed a statistically significant relationship, three showed excesses that did not reach statistical significance, and two "no relationship" (in one case without numbers being presented). Overall, no studies show a protective effect (such as would be expected in at least some studies if all findings were a product of random variation). Two of three studies involving objective prospective surveillance (tympanometry or insufflation otoscopy) showed statistically significant associations, and the third a nonsignificant excess of middle ear problems with ETS exposure. Both studies involving biomarkers of ETS exposure showed statistically significant relationships between exposure and outcome.