Tobacco Documents Online

CRITIQUE OF ENVIRONMENTAL TOBACCO SMOKE; A COMPENDIUM OF TECHNICAL INFORMATION Chapter 9: The Effects of Passive Sraoking and on Respiratory Illnesses in Children Day Care Prepared by: Ronald D. Hood, Ph.D. I- U c L I have.been a Professor of Bio:.ogy in the Cell, Molecular, and Developmental Biology Section, Department of Biology, The University of Alabama, since 1978. I hold a concurrent appointment as Adjunct Professor of Environmental Health Sciences in the School of Public Health, University of Alabama at Birmingham. I am also the Pr'.ncipal Associate in Ronald D. Hood and Associates, Toxicology Consultants. Since 1983, I have served as a Special Consultant to the EPA's Science Advisory Board. Since 1978, I have acted as a professional consultant in the areas of environmental, developmental and reproductive toxicology for a number of industrial clients and for government agencies, including the EPA, the Veterans Administration, and the Congressional Office of Technology Assessment. My curriculum vitae is attached. I have been asked to review "The Effects of Passive Smoking and Day Care on Respiratory Illnesses in Children," by Glen Bennett, which is Chapter Nine of an EPA draft compendium of technical literature on environmental tobacco smoke. Mr. Bennett's chapter on day care attendance and respiratory illnesses in children takes note of the many L

- 2 - L e problematic aspects of the epidemiologic studies.that have found a statistical association between parental smoking and the occurrence of certain respiratory diseases in children. Nevertheless, he appears to accept that an increased risk of such diseases is attributable to exposure to environmental tobacco smoke ("ETS"). He goes on to.suggest that smoking by day care workers -- a factor that he recognizes has not yet actually been studied -- may present a comparable risk. this basis, he proposes both that parents be alerted to a On possible danger in this area and that day care providers be educated about a possible compounding effect of ETS exposure and day care attendance. As an initial matter, I note that studies on parental smoking have been widely criticized. For example, a 1988 study by Rubin and Damus (14) re-examined the 30 extant studies on parental smoking and observed that "most studies had significant design problems that prevent reliance on their conclusions'." Additionally, a workshop sponsored by the NIH in 1983 (10) emphasized the importance of confounding factors for parental smoking studies; such factors include cross-infection in the home, proximity of the home to industry, general nutrition, and family access to medical care. Furthermore, at a recent international symposium on ETS at McGill University in Montreal, which I attended, Professor R. Witorsch (20) reached the following conclusions on the basis of a comprehensive review of the relevant literature: The association between parental smoking and increased incidence of respiratory symptoms L

3 and diseases in young children is provocative. However, the mechanism for this association remains unexplained. Among the possibilities to be considered are ETS, socioeconomic factors and effects of maternal smoking during pregnancy and/or lactation. The increased inconsistency of this association as the child ages is also unexplained. Amcng the possibilities to be considered for this apparent age-dependent change are changes in the susceptibility to or intensity of ETS exposure, inaccuracies in the data obtained from questionnaires (e.g. unvalidated clinical data and smoking misclassification) and confounding variables. L L Clearly a great deal of further research needs to be conducted before one could conclude with confidence that parental smoking itself is a causal factor for respiratory illnesses in early childhood. By the same token, it appears premature at best to employ an assumpticn about parental smoking as a basis for action by parents, day care providers or regulators with regard to smoking by day care workers -- particularly when our knowledge of the role of the day care environment itself is as yet so rudimentary. Ih the following review, I consider epidemiologic studies involving either parental smoking or day care attendance with regard to the only nexus between the two groups of studies, namely, the incidence of otitis media. Preliminary Observations Children in day care centers (DCCs) have significant health problems of various kinds, and sc it is important that any potential contributing factor be accurately identified. Children in DCCs have increased rates of many infectious diseases (2), including ones such as diarrhea, hepatitis, and

4 r Ii i meningitis that are by no stretch of the imagination related to ETS exposure. Children in DCCs also have an increased rate of colonization with antibiotic-resistant bacteria (3). There is, however, little evidence of high risk for respiratory illness, with the possible exception of Dtitis media. This infection problem is deemed moderate at::nost, and I agree that ". .. there is no need for a clarion call against day care or on behalf of major new federal or state regulation" in this situation (2). Some additional observations about DCC clientele are germane to the evaluations of studies conducted therein. It must be remembered (and controlled for in studies, if possible) that the reason for the child's being in a day care center may itself be an independent risk factor. Many mothers work because of financial need, and lower socioeconomic status is a known r•isk factor for infections. Lower socioeconomic status may also entail poorer nutrition, apartment (as opposed to single family home) dwelling, city (as opposed to suburban) living, and crowding, all of which are.pctential risk factors. It is also possible that DCC children may visit a doctor more often, perhaps because the mother becomes more concerned upon being presented suddenly with a sick child at the end of the day (as opposed to the mother gradually becoming aware of a child's illnesse while caring for it all day at home), or because DCC personnel influence the mother to seek medical attention for her child. Thus, at least some conditions may be more likely to be diagnosed in DCC children than in control L

- 5 - children, and this would exaggerate the ,apparent risk shown in -epidemiologic studies. Many of the available epidemiologic studies of DCC children have relied upon answers to questionnaires given by the day care provider. In the DCC the clzild is seen every day by someone knowledgeable about children, and this may lead to more diagnoses being identified •than would be the case with other types of care, again a factor that would falsely magnify the apparent risk. Additional potential confounders will be discussed later. Relationship to Respiratory Infections Otitis media is the only upper respiratory disease reported in the literature as being associated with ETS exposure, and it is the only one that ha,s been extensively studied in connection with DCC. Bennett'cited five such studies as being positive (see his references 3, 20, 21, 30, 34) and four as being negative (see his :ceferences 39, 45-47). It seems appropriate to review in greate:r detail the various studies that were cited and several addi:ional ones that were not. To facilitate comparison and evaluation, such standard. statistics for epidemiologic studies as -:he relative risk and its 95 per cent confidence intervals and probability (P) were calculated and are presented in Table 1. Table 1 summarizes the studies cited by Bennett of DCC or parental smoking and their association with otitis media or various of its parameters, plus some additional studies that were not cited by him. Four of the five studies

6 characterized by him as being positive*- reveal an increased incidence-of otitis media or its parameters that is statistically significant. Three of these are small case control studies, one was a retrospective cohort study, and there were no prospective studies. The Kraemer study (7) considered only smoking in the home and not day care use. That study found no effect with a single smoker or the smoking of fewer than three packs per day in the home, but did find an effect with two smokers or more than three packs -- factors not likely to prevail in a day care center, even if it were staffed with a smoker. The Said study's (15) reliability is open to serious question because it relied upon the subjects' remembering at age 10 to 20 what happene3 before age three, and it did not control for potential conEounders. The Black study (1) was the only one that controlled extensively for potential confounders or independent risk factors, but it had a probable detection bias. The diagnosis of "glue ear" in controls was based on parents' recounting of their physicians' diagnoses whereas subjects' diagnoses were confirmed by the investigators. If one considers that the disorder in many subjects was remarkably asymptomatic and that four controls had to be excluded because they had had surgery for glue ear, it becomes evident that the diagnosis was_more likely to be */ Bennett stated that there were six "positive" studies, hut listed only five. L_

Ld 7 missed in controls than in subjects. -This would have the effect of exaggerating the relative risk. All of these "positive" studies considered also (or in two cases, only) parental smoking, which in two studies was not associated with the difference noted with DCC, and in two instances the difference said to be assc-ciated with parental smoking was essentially similar to or greater than that associated with DCC. Of the four "negative" studies, only three concerned DCC; the remaining study (Van Cauwenberge, 17) dealt only with parental smoking and found no effect. Gne case control study (19) found a statistically significant increase in the incidence of otitis media in children with DCC, but the comparison was with much younger control subjects who may not have yet experienced their first bout of otitis -- predisposing to otitis media at the time of.examination -- which would exaggerate the relative risk. The others found no significant difference with DCC, and one (16) found with DCC a higher relative risk for fever or antibiotic use than for otitis media. Again, in this group of studies, those that considered parental smoking found no difference for that factor, including the study (19) that found a significant difference with DCC.

8 t I Especially noteworthy-among the-remaaining-studies are the lack of-control for parental smoking in most,*/ the almost universal lack of evident control for other confounders (Table 1), and the tendency for significant associations with DCC to be restricted to severe complications of otitis media. Yet here again a similar effect was not associated with parental smoking in those studies that included such an evaluation. One prospective study (4) found nearly a.50 per cent lower incidence of one to four'bouts of otitis-media.with DCC than with home care. This statistically highly significant difference ceased to exist only when five or more bouts of otitis media were considered. Then, the incidence was the same with DCC as with home care. Confounders Most studies of this issue have significant design problems (14). For reasons mentioned above, studies at day care centers may have a built-in bias toward reporting elevated ill.ness rates (2). The DCC studies presently fail to control for (though many studied the effect of) parental smoking, and there are as yet no studies that consider the smoking habits of the day care workers. All studies fail to control fully for confounders or independent risk factors, and */ It shouid be noted that looking for an effect of parental smoking is not the same as controlling for it. The latter requires matching of subjects and controls with regard to parental smoking or use of other procedures not possible to apply to the data as presented.

- 9 - many make no effort at all to control for even the most obvious ones (Table 1). Detection biases have already been mentioned. In many DCC studies there is also a substantial potential for misclassification bias caused by erroneous reporting of I parental smoking. Such misclassification biases are very important, because they can affect both index and control groups; occurrence in the former causes exaggeration, and in the latter causes dilution, of the apparent relative risk. Additional confounders that deserve consideration, but rarely have been controlled for in DCC studies, include outdoor air quality, home heating, air conditioning, and humidity. The failure of an experimental design to detect the same or a greater effect of parental smoking as compared with DCC attendance suggests very strongly that the DCC effect being measured is not due to smoking by DCC personnel. It is unlikely that one or even more smokers caring for a group of children in•a DCC setting would typically produce a concentration of ETS that would exceed that generated by a smoking, non-working parent. Such a parent would be likely to be in closer proximity to the child for larger portions of the week than would the day care worker. I suggest that what increase there might be in the incidence of otitis media and its parameters with DCC is merely a reflection of the increased incidence of various infections with crowding of children at very early ages, an observation that has been almost universal. If, as-Chapter 9 states, the total burden

of respiratory illness is actually similar for day care _ children and controls, any apparent differences in the incidences of otitis media probably reflect differences in either: (a) detection/diagnosis, or (b) the relative age at which children contract respiratory diseases. Conclusions Given the questionable conclusions of those studies that have found a statistical associaticn between parental smoking and childhood respiratory disease, one is not justified in extrapolating by means of a crude analogy from parental smoking to smoking by day care workers. In fact, day care attendance in general -- irrespective of whether day care workers smoke -- could well be a significant confounding factor in those parental smoking studies that have found a positive correlation. Furthermore, a review of those studies involving otitis media, which is both-the only upper respiratory disease that has been associated with ETS exposure and the only one that has been extensively studied in connection with day care attendance, strongly suggests that smoking by day care personnel is not a relevant factor. Finally, there are as yet no studies that actually consider smoking by the day care work-ers. This state of affairs points clearly to a need for conducting further research before the EPA disseminates a document such as the Bennett chapter. Otherwise, the public will very likely be misled to conclude that smoking in day care centers has already been established