© 2006 American Thoracic Society doi: 10.1164/rccm.200609-1337ED
Where There's Fire, There's SmokeUniversity of Washington School of Public Health and Community Medicine, Seattle, Washington For anyone who has been exposed to smoke from a large wildfire, it comes as no surprise to learn that it causes symptoms such as eye irritation and sore throat. One might reasonably ask whether something so obvious needs to be formally addressed by an epidemiologic study. Although we may know that wildfire smoke causes symptoms, we do not know the full range of symptoms or who is most susceptible, or whether more adverse effects occur. These are issues that can be addressed by an observational epidemiology study. In this issue of the Journal (pp. 1221–1228) Künzli and colleagues describe a study of reported respiratory symptoms, medication use, school absenteeism, and doctor visits in children in relation to wildfire smoke that affected large areas of Southern California in the fall of 2003 (1). The convergence of two unusual phenomena, the wildfire smoke and an ongoing study, made this investigation possible. Although wildfires are common, smoke from wildfires does not commonly affect large populations because wildfires typically occur in rural areas. Furthermore, the University of Southern California Children's Health Study (CHS), a longstanding cohort study of air pollution effects in 16 Southern California communities, was ongoing at the time of the wildfires (2, 3). The CHS investigators, then, were ideally placed to exploit this convergence. Because the window of opportunity was relatively short, a questionnaire was expeditiously distributed to active CHS participants to gather information from the children and parents on smoke exposure and health outcomes. Smoke exposure was graded after the fact according to the number of days that smoke was reportedly smelled indoors during the 2-wk peak period of smoke. As an objective measure of outdoor smoke, measured or estimated concentrations of inhalable particulate matter (PM10) were averaged over the peak 5-d period. What was found? Those children from households where the smell of smoke was reported indoors had more reported symptoms, medication use, doctor visits, and missed school, with an exposure–response relationship, than children in which there was no smell of smoke indoors. Although these findings are generally what we might have expected, we should not necessarily take them all at face value. The most obvious concern here is the potential for reporting bias, a form of differential information bias. In this case, those who were more symptomatic might have reported more exposure when in fact there was not, or vice versa. A study in which questionnaire responses were elicited in relation to an air pollution episode that was known to everyone, and where participants both graded exposure and reported symptoms, should alert one to the likelihood of reporting bias. The questionnaire itself focused attention on symptoms related to the smoke event, further adding to the concern. Additional concerns about selection bias and recall bias, as the authors note, are triggered by participation rates and time of completion of the questionnaire, respectively, which varied across communities. What clues do we have that reporting bias played a role? Estimates of effect were more likely to be increased and were generally substantially larger when exposure was based on within-community reported smell of indoor smoke than on community-wide ambient PM10 concentration. Most obvious, with PM10 concentration there was no effect on either physician visits or school absenteeism. Künzli and colleagues correctly point out that the exposure measure using measured concentrations and that using subjective assessment of duration of exposure "measure different domains of exposure," and also that there may have been substantial spatial variation in smoke concentrations within each community that would only have been picked up by the subjective measure (1). One might not, then, necessarily expect findings based on these different exposure measures to coincide. I began this editorial with an observation on the value of epidemiologic observations in defining the range and severity of exposure effects, even in settings where it is obvious that the exposure is having some effects. Because of concern over reporting bias as well as other biases in this study, we are left with some uncertainty as to the range and severity of wildfire smoke effects. We do not have the information we need to determine whether the subjective within-community exposure measure or the objective between-community measure should take precedence. Normally one would be inclined to give preference to an objective measure, but the findings using the subjective measure indicating that those who smelled smoke over a prolonged period were many times more likely to experience symptoms than those who smelled no smoke seem realistic. What do other studies of wildfire smoke effects have to tell us about effects on outcomes other than symptoms? The evidence is conflicting. An early study on wildfires in California in 1987 found effects on emergency room visits (4), but a study of bushfires in Australia showed no evidence of increased emergency room visits (5), and a recent study on wildfire smoke in Denver found no evidence of increased mortality (6). Studies on the 1997 fires in Southeast Asia arguably have more relevance to the California fires studied by Künzli and colleagues (1), at least in terms of the temporal pattern of urban population smoke exposures. Increases in medical clinic and emergency room visits, but no effects on hospitalizations or mortality, were seen in association with these fires in Singapore (7); an effect on mortality, however, was seen in Kuala Lumpur (8). Because of the consistent effect of the California wildfire smoke on symptoms reported by Künzli and colleagues (1), it is very plausible that the most susceptible of the exposed population would seek medical care. We might therefore feel justified in giving the nod to the findings based on the within-community subjective measure of exposure. Caution would be the wiser course. Epidemiologic studies are more valuable when we use them to attempt to reduce uncertainty, rather than to merely confirm what we expect. Although we know there are health impacts from exposure to wildfire smoke, we remain uncertain about the range and severity of these impacts. FOOTNOTES Conflict of Interest Statement: S.V. does not a financial relationship with a commercial entity that has an interest in the subject of this manuscript. REFERENCES
Related articles in AJRCCM:
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
