INTRODUCTION

Wildfire management involves many ecologic and social trade-offs (Gonzalez-Caban, 1997). Burning landscapes under controlled conditions to reduce fuel loads and, thereby, to minimize the frequency and severity of wildfires is a controversial type of landscape management. A particularly problematic aspect of this land management intervention concerns the health effects of smoke pollution from frequent, controlled landscape fires vs. severe smoke pollution from infrequent, uncontrolled wildfires (Schwela, 2001; Lewis and Corbett, 2002; Sim, 2002). It has been recently suggested that particulates derived from wood smoke might be more injurious to human health (Boman et al., 2003) than particulates derived from other sources that are also known to cause ill health (Vedal and McClellan, 2002; Katsouyanni, 2003). Despite this knowledge base, there is a rudimentary understanding of the health effects of wildfire smoke on human health (Table 1). Consequently, land managers and health professionals are operating in a policy vacuum with little to guide them beyond existing national air quality standards. For example, the U.S. Environmental Protection Agency (1998) produced an interim policy for the management of wildfire management that promoted the “thoughtful use of fire by all wildland owners and managers while mitigating the impacts of emissions on air quality and visibility.” In this context, the work of Johnston et al. (2002a) is very significant to the ongoing policy debate concerning fire management and air quality because, unlike all previous studies, they were able to study the effect of a wide range of smoke pollution levels at and well below national air quality standards.

Table 1 Summary of Previous Epidemiologic Studies of Wildfire Smoke and Human Health
Full size table

TEMPORAL PATTERNS IN ASTHMAAND WILDFIRE SMOKE

Johnston et al. (2002a) undertook a correlative analysis of atmospheric pollution and asthma in the isolated coastal tropical city of Darwin, situated in the vast savannas of the Australian monsoon tropics (Fig. 1). The study was conducted during the 7-month rain-free dry season in 2000. There are a number of advantages for the study of human health effects of wildfire smoke in Darwin: 1) there is no significant source of atmospheric air pollution other than particulates derived from wildfires (CSIRO Atmospheric Research, 2001); 2) during the dry season, the lower atmosphere is stable, with little convective mixing and a persistent inversion at approximately 3000 m (Kondo et al., 2003), allowing reliability of exposure measurement within the geographic area (CSIRO Atmospheric Research, 2001); 3) there is a small population of approximately 115,000 and a single major hospital that has systematic data collection systems in place; and 4) wildfires occur in the Darwin region throughout every dry season (e.g., Edwards et al., 2001), thereby providing a continuous background of smoke pollution with peaks and troughs over several months.

Figure 1
figure 1

The location of the Australian city of Darwin and other regional population centers. Also shown is the extent of the urban area within Darwin and the location of the university and the meteorology station where PM10 (atmospheric mass of particles 10 μg or less in aerodynamic diameter per cubic meter of air) levels were measured. PM10 loadings were continuously measured by using a tapered element oscillating mass balance at the university, and gravimetric mass loadings were measured at the meteorology station. Determinations from the two sites were highly correlated (r = 0.89; CSIRO Atmospheric Research, 2001).

Full size image

Smoke pollution was measured as PM10 (atmospheric mass of particles 10 μg or less in aerodynamic diameter per cubic meter of air) averaged over a 24-hour period at two locations within the city of Darwin; this provided a reliable measure of exposure of the city’s population (Fig. 1). For each day of the study, the number of asthma presentations to the emergency department of the Royal Darwin Hospital was determined by interrogating the electronic emergency department register, which was coded according to a subset of the International Classification of Diseases, version 9, codebook. The two codes used for asthma presentations were 493.00 (childhood asthma) and 493.9 (asthma not elsewhere classified), which was used to code all types of adult asthma.

Johnston et al. (2002a, b) demonstrated a significant relationship of asthma presentations with increasing PM10 concentrations by using negative binomial regression and controlling for potential confounders known to be coincident with fire activity and potentially influencing asthma presentations to hospital, including weekend and holiday periods and the incidence of influenza-like illness, a well-known precipitant of asthma (Cohen and Castro, 2003). They showed that when the PM10 data were analyzed as a continuous variable, there was a significant increase in asthma presentations with each 10 μg/m3 increase in PM10. However, when the PM10 data were aggregated into 10 μg/m3 PM10 classes, the modeling revealed a threshold at 40 μg/m3, where the risk of asthma increased 2.6 times relative to the baseline category of less than 10 μg/m3 (Table 2).

Table 2 Rate Ratio and Confidence Intervals of Asthma Presentations and Exposure Levels of PM10 after a Negative Binomial Regression
Full size table

Time series has been the predominant and orthodox approach for longitudinal studies of air pollution at a number of urban loci and is used to overcome the effects of autocorrelation due to seasonality, weather fluctuations, and long-term trends or cycles (Jalaludin et al., 2002). Reanalysis of Johnston and associates’ original data to test for temporal autocorrelation (by using autoregressive integrated moving average autoregressive integrated moving average (ARIMA) modeling) indicated that their original analysis was robust: there was no evidence of temporal autocorrelation in asthma presentations or any obvious trend or cycle in the data (Fig. 2).

Figure 2
figure 2

Daily asthma presentations to the emergency department of the Royal Darwin Hospital and 24-hour mean concentration of atmospheric particles (PM10) measured at two locations in the city of Darwin from April 1 to October 31, 2000. With autoregressive integrated moving average (ARIMA) analysis, the highest partial autocorrelation coefficient for asthma presentations was 0.13 at 1 day, with no trends or cycles in the data. The mean daily PM10 for the entire study period was 20.84 μg/m3, and it ranged from 2.0 to 70.0 μg/m3. The observed PM10 levels are representative of the regional setting, because Vanderzalm et al. (2003) found similar levels of carbon particles at Jabiru (Fig. 1) in the dry season of 1996.

Full size image

THE NEXUS BETWEEN HUMAN ANDL ANDSCAPE HEALTH

A manifestation of global environmental change is the worldwide increase in the severity and frequency of wildfires (Schwela, 2001). Because of the effects of wildfires on human populations, there is an increasing need to consider this problem in a holistic manner, including understanding the effect of both wildfires and prescribed fires on human health and landscape ecology. Demonstrating a health effect of wildfires presents numerous practical and technical challenges because of the infrequency of fire events and the large number of potential confounders that make it difficult to distinguish the epidemiologic “signal” from the background “noise.”

The Johnston et al. (2002a) study represents an advance in this field because it examined temporal variation in smoke pollution over 7 months of continuous exposure rather than using a more usual retrospective design comparing an unexpected fire event with an “equivalent” period without wildfire (Table 1). The finding that asthma hospital presentation more than doubled at more than 40 μg/m3 PM10 is potentially of great significance for fire management, because this is below currently accepted Australian air quality standards (i.e., <50 μg/m3 PM10). Furthermore, this threshold was rarely exceeded by fires in the early and mid dry season, most of which were deliberately lit to reduce fuel loads. This observation lends support to the idea that well-managed controlled burning programs may, for asthma, have negligible public health effects relative to uncontrolled fires that cause severe pollution (Table 1).

It must be stressed that the 40 μg/m3 PM10 threshold detected by Johnston et al. (2002a) may be a statistical anomaly, given that it is inconsistent with previous studies of health effects of particulates. To resolve this question, we are exploiting the unique opportunity presented by Darwin to undertake a more comprehensive epidemiologic study of the associations between wildfire smoke and health. This study includes measures of particulate pollution measured as PM10 and PM2.5, spores and pollens, meteorologic conditions, day of the week, and school holiday time periods. In contrast to previous study designs, we are examining multiple health outcomes, including tracking the response of a cohort of patients with asthma to varying levels of wildfire smoke exposure by recording their daily symptoms, medication use, and health service attendances for asthma. Additionally, for the city of Darwin, we are tracking hospital attendances for heart and lung diseases (including asthma, chronic obstructive pulmonary disease, and ischemic heart disease) and family physician presentations for rhinitis and flu-like illness. To understand the interactive effect of meteorologic conditions and the geographic location of fires on smoke pollution levels in Darwin, we are undertaking landscape ecology studies by using moderate resolution imaging spectroradiometer satellite imagery to locate active wildfires and by using a geographic information system and dynamic atmospheric modeling to map smoke dispersion from them (Draxler and Hess, 1998). Such ambitious joint landscape ecology and epidemiologic perspectives are crucial in the quest for sustainable fire management practices for a variety of reasons, including building a bridge between the presently segregated professions of fire management and public health; providing urban dwellers with a direct stake in fire management; and providing specific, evidence-based, and measurable air quality targets for fire managers.