Annals of Occupational Hygiene Advance Access originally published online on May 31, 2008
Annals of Occupational Hygiene 2008 52(6):497-508; doi:10.1093/annhyg/men027
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Occupational PAH Exposures during Prescribed Pile Burns
1 Department of Chemistry and Biochemistry, Northern Arizona University, Flagstaff, AZ, USA
2 Community, Environment and Policy, Mel and Enid Zuckerman College of Public Health, University of Arizona, 1295 North Martin Avenue, PO Box 245163, Tucson, AZ 85724, USA
3 Department of Chemistry and Biochemistry, Arizona State University, Tempe, AZ, USA
* Author to whom correspondence should be addressed. Tel: +1-520-626-4918; fax: +1-520-626-8009; e-mail: jburgess{at}u.arizona.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
Wildland firefighters are exposed to particulate matter and gases containing polycyclic aromatic hydrocarbons (PAHs), many of which are known carcinogens. Our objective was to evaluate the extent of firefighter exposure to particulate and PAHs during prescribed pile burns of mainly ponderosa pine slash and determine whether these exposures were correlated with changes in urinary 1-hydroxypyrene (1-HP), a PAH metabolite. Personal and area sampling for particulate and PAH exposures were conducted on the White Mountain Apache Tribe reservation, working with 21 Bureau of Indian Affairs/Fort Apache Agency wildland firefighters during the fall of 2006. Urine samples were collected pre- and post-exposure and pulmonary function was measured. Personal PAH exposures were detectable for only 3 of 16 PAHs analyzed: naphthalene, phenanthrene, and fluorene, all of which were identified only in vapor-phase samples. Condensed-phase PAHs were detected in PM2.5 area samples (20 of 21 PAHs analyzed were detected, all but naphthalene) at concentrations below 1 µg m–3. The total PAH/PM2.5 mass fractions were roughly a factor of two higher during smoldering (1.06 ± 0.15) than ignition (0.55 ± 0.04 µg mg–1). There were no significant changes in urinary 1-HP or pulmonary function following exposure to pile burning. In summary, PAH exposures were low in pile burns, and urinary testing for a PAH metabolite failed to show a significant difference between baseline and post-exposure measurements.
Keywords: particulate matter • polycyclic aromatic hydrocarbons • prescribed burn • respirable dust • wildland firefighter • wood smoke
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
The risk of catastrophic wildfire in the USA is expected to increase in the next decade. In 2006 alone, 96 385 wildland fires burned over 3 995 762 ha (9 873 745 acres), making 2006 a record year for hectares burned (National Interagency Fire Center, 2008). To reduce the threat of wildfire, federal, tribal and state land managers have agreed to increase the use of prescribed fire (US EPA, 1998). A prescribed fire is a low-intensity burn used to remove years of built-up fuels and restore forest ecosystems to a healthier state. Two common types of prescribed fire are broadcast burns, where fire moves quickly through a 40–200 ha (100–500 acre) region consuming largely small diameter fuels (e.g. dead pine needles and branches) and pile burns, where slash and logging debris are piled before burning to increase fire residence time.
By intentionally burning these fuels, the risk of wildfire is greatly reduced. However, the wood smoke generated during prescribed fire may expose firefighters to occupational health risks, including cancer. For example, increased cancer rates including cancers of the brain, leukemia, non-Hodgkin's lymphoma and both bladder and kidney cancers and less often cancers of the nasopharynx, prostate, colon and skin have been reported among structural firefighters (Sama et al., 1990; Grimes et al., 1991; Demers et al., 1992; Guidotti, 1993; Aronson et al., 1994; Burnett et al., 1994; Tornling et al., 1994; Delahunt et al., 1995). Whether these or other health effects are associated with wood smoke from prescribed fires remains poorly understood.
Wood smoke is a well-known source of particulate matter with often >70% of its mass classifiable as PM2.5 or particles with aerodynamic diameters 
2.5 µm (Lighty et al., 1995; US EPA, 1996; NWCG, 2001). In fire-generated particles, the largest fraction is typically <1.0 µm (Kleeman et al., 1999). Such particles are small enough to reach the alveoli in the lungs and have been linked to pulmonary disease, morbidity and even death (Dockery et al., 1993; Schwartz, 1994; Hoek et al., 1998).
The chemical composition of particulate matter also influences its exposure risk. Wood smoke particulate is a complex mixture of ionic, inorganic and organic species with organic species comprising the largest mass fraction (>60%). To date, >300 organic compounds have been identified in wood smoke particulate, but >1000 are expected to be present (e.g. see bin Abas et al., 1995; Simoneit et al., 1993; Fine et al., 2002; Hays et al., 2002; Fine et al., 2004). Among the identified organic compounds, anhydrosugars (e.g., levoglucosan), resin acids (natural compounds found in plant materials), alkenoic and alkanoic acids and diterpenoids (in natural and heat-altered forms) are generally most abundant. Wood smoke particles also contain polycyclic aromatic hydrocarbons (PAHs). Although PAHs comprise only a small fraction of the particulate mass (<5%), they are important because of their potential toxicity. PAHs are procarcinogens, requiring metabolic activation to electrophiles (Penning et al., 1999), but once activated, they have been linked to diseases such as lung and bladder cancer (Ronneberg et al., 1999; Boström et al., 2002; Gaertner and Theriault, 2002).
PAH levels in wood smoke vary with factors such as fuel type (e.g. oak or pine), fuel source (e.g. logs, deadwood, litter or other decomposing organic matter), moisture content and combustion temperature (e.g. flaming or smoldering conditions), to name only a few. For example, burning wood in a residential fireplace appears to generate more particle-phase PAHs than burning wood in an open-air campfire (Simoneit et al., 2000). Also, in fireplace combustion studies, pinyon pine produced higher PAH levels than ponderosa pine (Oros and Simoneit, 2001; Fine et al., 2002).
Despite their importance, few studies exist on PAH levels in wood smoke during prescribed fire or their impacts on wildland firefighters. Studies of prescribed burns in California (Materna et al., 1992; Reh et al., 1994) suggest that PAH levels may be higher in gas-phase than in particle-phase smoke. Low particle-phase PAH levels were also observed at four prescribed fires in Georgia (Lee et al., 2005); however, PM2.5 samples were collected 21–26 km downwind of the burn site, and personal exposures were not measured. For structural firefighters, PAH exposures have been confirmed in measurements of 1-hydroxypyrene (1-HP), a urinary PAH metabolite, following firefighting activities (Moen and Ovebro, 1997; Caux et al., 2002); however, comparable studies on wildland firefighters do not exist.
The US Department of Agriculture has recommended additional characterization of the chemical composition of particulate matter in smoke from wildland fires (Sharkey, 1997). Given the expected increase of wildland fires over the next decade and concerns for PAH exposures and their associated health effects, an evaluation of exposure risk and measurements of short-term health effects are warranted. In this work, personal exposures to wood smoke from pile burning in Native American wildland firefighters were assessed. Native Americans reportedly account for half the firefighters on the front lines of wildfires in the USA, although they make up roughly 1% of the country's population (Hardin, 2004). To our knowledge, studies involving Native American firefighters have not been reported. Nor is it known if Native American firefighters are more or less susceptible to wood smoke exposures than other populations.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
This project was approved by the Institutional Review Boards of Northern Arizona University and the University of Arizona, the US Department of the Interior Bureau of Indian Affairs, the White Mountain Apache Health Advisory Board and the White Mountain Apache Tribal Council. The study methods had three main components: (i) human subject recruitment and health screening before and after prescribed pile burning, (ii) personal exposure monitoring during pile burns and (iii) area monitoring during pile burns.
Subject recruitment and evaluation
The study was conducted on the White Mountain Apache Reservation, which encompasses 647 500 ha (1.6 million acres) in east central Arizona, in Apache, Gila and Navajo Counties, and has >12 000 members. The White Mountain Apache Tribe (WMAT) has a proud tradition of wildland firefighting. There are 
400 WMAT firefighters, of whom roughly 50 participate in prescribed burns. Firefighters chosen by the Fort Apache Agency for prescribed burning activities were informed of the study protocols and risks. Those choosing to join the study underwent baseline testing including a health/occupational questionnaire and spirometry. Only those who believed they would continue to work as an Apache wildland firefighter during the entire first year of the study were included. Subjects were excluded from the study if they participated in wildland firefighting or prescribed burning within 5 days of baseline testing.
The pre-season questionnaire was based on a standardized questionnaire (ATS-DLD-78, Ferris, 1978) and a validated symptom scale (Wasserfallen et al., 1997). The modified questionnaire collected information on chronic respiratory conditions, tobacco history, symptom history over the past week and occupational history. The pre-season questionnaire included questions on the use of coal or coke, wood, gas, fuel oil, kerosene or electricity separately for heating and cooking. The post-fire questionnaire focused on temporal exposures. Additional questions were included in both questionnaires to assess non-occupational exposure to PAHs, including consumption of grilled meat.
Burn sites
The study took place on the White Mountain Apache Reservation (Fig. 1) during the fall 2006 prescribed fire season. The reservation is situated in the Apache–Sitgreaves National Forest of Arizona, a predominately ponderosa pine forest at the elevations studied (2200 m). In total, five prescribed burns (all pile burns) were conducted, two at the Hon-Dah site (8 November and 6 December), two at the Maverick site (10 November and 7 December) and one at the White River site (30 November).
|
Personal exposures were monitored on Day 1 (ignition/flaming) of each burn, when smoke levels and human exposures were expected to be greatest. Personal exposures were also monitored on one ‘no-burn’ day, when weather conditions prevented pile ignitions. Area sampling was conducted on Day 1 (ignition/flaming) and Day 2 (smoldering) of two burns (8 and 10 November) in order to compare PAH levels under both flaming and smoldering conditions. Daytime temperatures were between 9.1 and 20.4°C, humidity was <32% and wind velocities were between 4.5 and 9.0 m s–1 (10–20 m.p.h.). Between 24 and 202 ha (60–500 acres) were burned during each event (average = 95 ha), with 300–2650 piles per event (average = 1308). Ponderosa pine slash (15 cm and less diameter tops and limbs, classified as 10-, 100- and 1000-h timelag fuels, referring to the time needed for fuel moisture to reach equilibrium with the environment) was the primary pile material, with other materials including oak, Douglas fir and other conifers.
The job categories for prescribed burning included igniter, monitor and patroller. To ignite the piles for the prescribed burns, a 3:1 diesel/unleaded gasoline mixture was used. (A red dye was added to the diesel fuel to signal its use for a non-commercial purpose.) The igniters traveled from pile to pile, starting in one corner of the assigned area and then igniting subsequent piles in an order that minimized smoke exposure. The monitors remained in one area and were responsible for containment of the line in that small area, using standard equipment (hand tools) to contain the fires. The patrollers walked up and down the line over a wide area to monitor the pile fires and to check for spot fires along their line. On Day 2, firefighters were not personally monitored for dust or PAHs. However, the patrollers continued to patrol the entire area to contain the smoldering fires and check for spot fires.
Personal air sampling
Three different types of personal exposure samples were collected on the firefighters: respirable dust (National Institute for Occupational Safety and Health (NIOSH) 0600, gravimetric), total dust (NIOSH 0500, gravimetric) and PAH scan (NIOSH 5506, high performance liquid chromatography (HPLC) with ultraviolet spectroscopy). The gravimetric methods were modified by sampling onto pre-weighed polytetrafluoroethylene (PTFE) filters rather than polyvinylchloride (PVC) filters to accommodate future PAH analyses of these samples. Respirable sampling used SKC aluminum cyclones, operated at 2.5 l.p.m. to achieve a 4-µm cut point, with pre-weighed 5-µm pore size PTFE filters in 37-mm polystyrene cassettes. Total dust sampling used 37-mm closed-face polystyrene cassettes with pre-weighted 5-µm pore size PTFE filters (2.5 l.p.m.). PAH samples used a 37-mm closed-face cassette with a 2-µm pore size PTFE filter followed by a 100/50-mg XAD-2 sorbent tube (2 l.p.m.). PAH samples were analyzed for a panel of 16 analytes, distinguishing between PAHs in the particle and vapor phases.
The 37-mm cassettes were used instead of inhalable samplers for several reasons: (i) based on previous measurements, the smoke particulate was expected to be fine, making the choice of sampler less critical; (ii) the cassettes were easier to deploy, encouraging more widespread use by firefighters and (iii) the NIOSH method cassettes were more consistent with the NIOSH method PAH samplers, also used in the study.
Air was drawn through the sample media using SKC Aircheck samplers (SKC Inc., Eighty Four, PA, USA). Sample pumps were calibrated with representative sampling media in line, both before and after each sampling event. A Bios DryCal DC-Lite (Bios International Corporation, Butler, NJ, USA) was used to verify flow rates.
The 21 firefighters were randomly assigned to wear one of three combinations of samplers: total dust only (n = 1), total and respirable dust (n = 8, but 3 were worn on a non-burn day when weather prohibited pile ignition) and total dust and PAH (n = 12). The total and respirable dust measurements were used to test our assumption that smoke particles were largely in the respirable range. When two samplers were assigned to a firefighter, two separate and complete sampling trains were worn by the firefighter. Sample pumps were clipped on opposite collar points of the firefighters' Numax outerwear, and sample media were positioned in the firefighters' breathing zones and within 23 cm (9 inches) of the mouth, for the duration of the shift. Tygon tubing and luer adapters were used to connect sample media to the sample pumps. On heavy smoke days, media were changed mid-shift to prevent overloading. Firefighters wore samplers during the entire work shift, generally lasting 8 h. All exposure estimates are presented as time-weighted exposures over the working shift.
Total particulate, respirable particulate and PAH concentrations were analyzed at Aerotech Environmental Laboratories (Phoenix, AZ, USA), accredited by AIHA IHLAP for all analyses requested. Sixteen PAHs were included in the scan: acenaphthene, acenaphthylene, anthracene, benz[a]anthracene, benzo[a]pyrene, benzo[b]fluoranthene, benzo[g,h,i]perylene, benzo[k]fluoranthene, chrysene, dibenz[a,h]anthracene, fluoranthene, fluorene, indeno[1,2,3-cd]pyrene, naphthalene, phenathrene and pyrene. Field blanks were collected for each sample type to evaluate contamination in the field or in transport to the laboratory; no sample contamination was identified.
Area sampling
Ambient PM2.5 samples were collected using a US Environmental Protection Agency-approved speciation sampler (SuperSASS, Met One Instruments, Inc., Grants Pass, OR, USA). The monitor has four independent channels that can be operated simultaneously. Each channel was equipped with its own sharp-cut cyclone (to remove particles >2.5 µm), automatic flow controller (flow rate 6.7 ± 0.5 l.p.m.), filter holder and filter media (47 mm). Channels 1 and 3 supported PTFE filters (2 µm, Whatman) for gravimetric analysis; channels 2 and 4 supported pre-baked quartz-fiber filters (PallFlex Tissuquartz) for PAH analysis. All four sampling filters were followed by a pre-baked quartz-fiber backup filter to test for filter adsorption artifacts (i.e. volatile PAHs that had sorbed onto the quartz-fiber filters). Naphthalene (two fused rings, MW = 128) was not detected on backup or sampling filters, suggesting that it was sufficiently volatile to be stripped off of both filters. Phenanthrene and anthracene (three fused rings, MW = 178) were detected in roughly equal amounts (>70%) on both filters, suggesting that they are primarily gas-phase PAHs. Reported concentrations were not corrected for these artifacts.
Area sampling was conducted on Day 1 (ignition/flaming) and Day 2 (smoldering) of two prescribed burns. Because no electrical source was available at the burn sites, the sampler was battery operated. On ignition days, for safety reasons, the sampler was placed just outside of the burn area, in a region expected to receive high levels of smoke. The sampler was started within 1 h of ignition; collection times were 2.5 and 2.8 h on the first and second ignition days, respectively. On Day 2, the sampler was placed inside the burn area, once in an open area and once near a smoldering slash pile. The collection time was 3 h on both days. For all area samples, once positioned, the monitor could not be moved during that burn event; hence, smoke intensities varied as the wind shifted during sampling. Filters were transported to and from the burn site in a cooler (<10°C). Quartz-fiber filters were stored at –20°C until analyzed; PTFE filters were stored at room temperature. Two field blanks, one PTFE and one quartz fiber, were collected during each burn event.
PTFE filters were analyzed gravimetrically (Chester LabNet, Tigard, OR, USA). Following conditioning in a control chamber (24 h, 20–25°C, 30–40% relative humidity), filters were weighed (±0.001 mg) pre- and post-exposure to determine PM2.5 mass. Area PM2.5 concentrations were obtained by dividing particulate mass by the volume of air pulled through the filter (mg m–3). Average concentrations (±standard deviation) from the two co-collected PTFE filters are reported. PTFE field blanks increased in mass by 28 µg (<3.0% PM2.5 mass), indicating that filter contamination was minimal.
Quartz-fiber filters were analyzed for PAH concentrations using gas chromatography/mass spectrometry (GC/MS). The procedure is well established and described elsewhere (Brown et al., 2002; Engling et al., 2006). In brief, filters were spiked with a series of deuterated internal standards (naphthalene-d8; pyrene-d10; chrysene-d12 and perylene-d12; all Cambridge Isotope Laboratories, Andover, MA, USA) and then extracted three times (3 x 25 ml) with methylene chloride (Fisher, Optima grade) under sonication. To maximize PAH concentrations, extracts from one and one-half co-collected quartz-fiber filters (channels 2 and 4) were combined into a single sample. The sample was reduced to 250 µl under a gentle flux of purified nitrogen. A 1-µl aliquot of the sample was injected into the GC/MS system (Agilent 6890/5973inert). PAHs were identified and quantified by comparison with retention times and peak areas of 21 PAH calibration standards. Field blanks showed no evidence of PAH contamination.
Particle-phase PAH concentrations are reported in three ways: (i) PAH mass per volume of air sampled (µg m–3), (ii) PAH mass per PM2.5 mass (µg PAH per mg PM2.5) and (iii) PAH mass per organic carbon mass (µg PAH per mg OC). PM2.5 quartz-fiber filter masses were determined by averaging the masses of the two co-collected PTFE filters and correcting for air volume. Errors are estimated to be ±15%. Organic carbon masses were determined by thermal optical transmission methods (Sunset Laboratory, Inc., Tigard, OR, USA) (Birch and Cary, 1996) using one-quarter of one quartz-fiber filter per burn event. The organic carbon mass (µg OC cm–2) was multiplied by the appropriate filter deposit area and corrected for air volume. Errors are estimated to be ±15%.
Biological sampling
Urine samples were collected pre-exposure (first morning void), at the end of the prescribed-burn shift, and the morning following prescribed burning (first morning void). The two post-exposure sample collections were needed to address the variability of elimination rates for urinary 1-HP. Previous studies indicate that maximum elimination occurs 6–7 h following PAH ingestion and 10–15 h following dermal exposure (Buckley and Lioy, 1992; Viau and Vyskocil, 1995). Urinary 1-HP samples were analyzed using HPLC with fluorescence detection (Carmella et al., 2004). Spirometry testing was performed using an EasyOneTM spirometer (ndd Medical Technologies, Chelmsford, MA, USA) following American Thoracic Society standards (ATS, 1995) both pre-exposure and at the end of the prescribed-burn shift. Sputum samples were also collected but are not reported in this work.
Statistical analysis
Exposure data were analyzed for mean and variability, in aggregate and by job category. Exposure data were tested for normality and lognormality using Shapiro–Wilk tests; the underlying distribution of the exposures was investigated, and upper tolerance tests of the 95th percentile were conducted to determine the upper limits of personal exposures. Simple linear regression was conducted on firefighters with matched samplers, total versus respirable dusts and total dust versus PAH to determine the relationship between exposure metrics. Wilcoxon sign-rank tests were performed to evaluate for potential changes in firefighters’ lung function (as indicated by forced expiratory volume in one second (FEV1) and forced vital capacity (FVC) levels) after working a controlled burn shift. For 1-HP data, concentrations below the detection limit were imputed a value of half the lower bound of detection. For the three time points of measurement (pre, post and morning after shift), a total of five measurements out of 60 were below detection. Sign-rank tests were again used to detect potential cross-shift changes in 1-HP concentration between pre- and post-exposure, as well as the next morning. To determine the potential relationships between environmental exposures (i.e. total dust and naphthalene) and changes in biological measurements (1-HP, FEV1 and FVC), Spearman correlations were estimated. Stata version 9.2 (College Station, TX, USA) was used for all statistical analyses.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
Personal sampling
Twenty of the 21 firefighters participating in the study were from the WMAT. They were predominantly male and averaged 41 years of age (Table 1). One subject who started to smoke between pre- and post-exposure measurements was excluded from subsequent pulmonary function and urinary 1-HP comparisons. In addition, two subjects did not have valid post-exposure measurements and were excluded from pre–post spirometry comparisons.
|
In this study, individual firefighters served multiple roles during the monitoring period. For example, of the four burn bosses, three also served as igniters and one also served as a patroller. Table 2 indicates the amount of overlap between job categories, with the numbers on the diagonal indicating the total number of workers who ever performed that role during the day (ever did), with other cells indicating how many workers also performed other roles.
|
In Table 3, total and respirable dust exposure data are grouped according to job category (ever did) and in aggregate. Twenty-one firefighters were monitored for total dust exposures during 5 days of active pile burning. Full-shift exposures ranged from <0.01 to 6.2 mg m–3 (mean = 1.01, SD = 1.37 mg m–3). Six additional gravimetric samples were collected on a no-burn day, when weather conditions prevented pile ignitions; all concentrations on this day were <0.4 mg m–3 and were not included in subsequent analyses. The underlying distribution was identified as lognormally distributed (Shapiro-Wilk W = 0.993). In evaluating exposures for the group, where each person was sampled exactly one time during the study, the upper 95% confidence limit for the 95th percentile was 8.6 mg m–3. Persons who performed ‘monitoring’ duties during the days sampled had a higher variability of exposure measurements.
|
One paired total and respirable dust exposure measurement was also collected on each active burn day to test our hypothesis that smoke exposures were primarily respirable (Table 3). Individual respirable exposures ranged from non-detectable (<0.13) to 1.25 mg m–3. The ratio of total to respirable dust ranged from 1.0 to 6.1, indicating that one firefighter was exposed to predominantly respirable dust while others were exposed to particles larger than 4 µm. The source of the larger particles was most likely the dust created by fire vehicles traveling back and forth on unpaved roads in the fire area. Simple linear regression failed to find a predictable relationship between the total and respirable dust exposure (R2 = 0.16), indicating the size distribution of particles varied among exposures.
Twelve firefighters were randomly assigned to wear samplers for PAHs. Tandem sample media were used, per NIOSH method 5506, and a time-weighted average (TWA) was determined for both vapor- and particle-phase PAH exposures. In the vapor phase, three of the 16 PAHs analyzed were detected: naphthalene, phenanthrene and fluorene (Table 4); the remaining 13 PAHs were below detection limits (0.4–14 µg m–3, depending on the analyte and sample volume). In the particle phase, no PAHs were detected on any sample.
|
The ACGIH (2007) occupational exposure limit for naphthalene is 50 000 µg m–3 (10 000 p.p.b.), and firefighter exposures at the upper tail of the underlying distribution curve of exposures were orders of magnitude below this value (mean = 6.2 µg m–3). The firefighting exposures were also in the low range compared to naphthalene exposures in other industries (8 h TWA, µg m–3): <1 in road construction; 2.2 in tar distillation (low temperature); 9 and 45 in foundry work; 32, 49 and 56 in coke oven work; 215 in tar distillation (high temperature) and 658 in timber impregnation (Unwin et al., 2006). Using simple linear regression on paired exposure data, we found that measurements of total dust were not correlated with exposure to any PAH component (n = 12, R2
0.01), and therefore total dust measurements could not be used to predict PAH exposure.
Area sampling
Area PM2.5 concentrations were measured during ignition/flaming (Day 1) and smoldering (Day 2) for two burns (Table 5). During ignition, one value (6.98 mg m–3) was considerably higher than the other (1.05 mg m–3). The higher value resulted from a more favorable placement of the PM2.5 monitor in the smoke plume and a relatively constant wind direction. On this day (8 November), the monitor (viewed from 25 m away) was visible less than half of the time due to heavy smoke levels. Similarly, the lower value resulted from a less favorable placement of the monitor and variable winds. On this day (10 November), the monitor remained fully visible throughout the sampling period, and winds often carried the smoke away from the monitor. Visible smoke levels decreased substantially during smoldering (Day 2). At Hon-Dah on Day 2, the sampler was placed in an open area, not near any smoldering piles, resulting in a lower PM2.5 concentration (0.90 mg m–3). At Maverick on Day 2, the sampler was placed just a few meters downwind of a smoldering pile, resulting in a higher value (2.89 mg m–3).
|
Area particle-phase PAH concentrations are reported in Table 6. Of the 21 PAHs analyzed, all were detected in the PM2.5 particles except for naphthalene. Naphthalene, because of its low molecular weight (128), is highly volatile and partitions nearly exclusively to the vapor phase (Harner and Bidleman, 1998; Kaupp and McLachlan, 1999), suggesting why the personal samplers detected naphthalene in the vapor phase (Table 4) but not the particle phase. Total PAH concentrations were 0.61 and 3.65 µg m–3 during ignition/flaming and 0.86 and 3.38 µg m–3 during smoldering. In each case, the higher value was measured on a day when the monitor was positioned more favorably in the smoke plume. With one exception (phenanthrene on Day 2 at Maverick), the highest individual PAH concentrations were observed on Day 1 at Hon-Dah; pyrene had the highest concentration (0.463 µg m–3), followed by fluoranthene, fluorene, benz[a]anthracene and chrysene + triphenylene (all >0.33 µg m–3).
|
Even though the mass concentrations of PM2.5 varied as a function of sampler position and wind, comparison of the PAH mass fractions on the collected particles is useful. Average PAH/PM2.5 mass fractions (µg PAH per mg PM2.5) during ignition/flaming and smoldering are plotted in Fig. 2, grouped by PAHs of the same molecular weight. (Naphthalene and fluorene were omitted due to low concentrations.) The lowest molecular weight PAHs (178) had the largest standard deviations, most likely due to their greater volatility. In all cases, the mass fractions were higher during smoldering. The total PAH/PM2.5 mass fractions (summing over all 21 PAHs) are reported in Table 6. Similar to the total PAH count data, the mass fractions were higher during smoldering (1.06 ± 0.15) than ignition (0.55 ± 0.04 µg mg–1).
|
Biological monitoring
Among the 18 firefighters with valid pre- and post-exposure measurements, no significant changes were found comparing baseline to post-exposure urinary 1-HP samples (Table 7). After excluding three subjects who ate grilled foods the day before or the day of baseline or post-exposure urinary testing, the differences between baseline versus end-of-shift and baseline versus next-AM mean urinary 1-HP concentrations remained non-significant. The measured urinary 1-HP measurements were all substantially less than the recommended American Conference of Governmental Industrial Hygienists Biological Exposure Index of 1 µg l–1 (ACGIH, 2007). There was no significant difference in urinary 1-HP concentration by type of fuel used for household heating and cooking, although the small sample size limited the study power in this regard. No significant changes in pulmonary function were noted from baseline to post-prescribed burn testing [3.66 ± 0.88 and 3.61 ± 0.88 l for FEV1 (P = 0.283) and 4.37 ± 1.05 and 4.40 ± 1.05 l for FVC (P = 0.693)]. However, due to the heavy fire season in 2006, seven of our firefighters had worked on other fires prior to this study. Hence, the baseline pulmonary function testing can only be considered pre-firefighting season for 14 of the 21 participating firefighters, although all subjects had at least 5 wood smoke-free days prior to baseline testing.
|
When comparing the exposure and biological measurements, no significant correlations (Spearman) were found after normalizing variable distributions. Total dust exposure was not correlated with change in baseline versus end-of-shift urinary 1-HP (P = 0.217). Furthermore, personal naphthalene exposure was not correlated with change in baseline versus end-of-shift 1-HP (P = 0.509). After excluding the three individuals who had eaten grilled foods the day of or the day prior to urinary testing, correlations of total dust and naphthalene to changes in 1-HP remained non-significant. In addition, personal naphthalene exposure was not correlated with change in FEV1 (P = 0.170), and total dust exposure was not correlated with change in FEV1 (P = 0.807) or FVC (P = 0.570).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
We identified relatively low particulate and PAH exposures to White Mountain Apache wildland firefighters during prescribed pile burns, and these exposures did not result in acute increases in urinary 1-HP levels. Based on discussions with the firefighters, the pile burns monitored during this study were representative of pile burns in general for the Fort Apache Agency. The firefighters evaluated in this study were able to change their locations to avoid areas of heavier smoke exposure, an observation supported by the finding that the PM2.5 area monitor measured higher concentrations than the personal respirable dust (PM4) monitors.
In the pile burns assessed in our study, mean TWA total particulate exposures were below 10 mg m–3. On average, igniters had the lowest exposures, consistent with the trend reported by Reinhardt et al. (2000). Alternatively, firefighters who reported monitoring as a function any time during their shift had exposures that were greater and more variable; hence, monitors may have the greatest potential exposures. For the Fort Apache Agency, monitors hold the fire line in one small area while patrollers hold the line up and down the road. Thus, the role of monitor appears consistent with what Reinhardt et al. (2000) called ‘holding the line’, the activity for which exposures were also the greatest and most variable in their Pacific Northwest study.
In the vapor phase, the personal monitors identified three PAHs. These vapor-phase contaminants were the lowest molecular weight compounds in the PAH scan, indicating that they may have been stripped off of particles collected on the PTFE filter immediately upstream of the sorbent tube. In the particle phase, the personal monitors did not detect any PAHs. This is in contrast to the area monitor, which detected 20 particle-phase PAHs. There are several reasons for this difference. First, the area monitor had lower detection limits than the personal monitors. Second, the area monitor collected larger samples due to a higher flow rate, a larger filter size and the ability to combine 1.5 filters into a single sample. As a result, the area monitor sampled between 1.5 and 1.8 m3 of air compared to 1.0 m3 of air for the personal monitors. Also, on one occasion (8 November), the area monitor spent over 2 h in a very smoky area, an area that firefighters wearing personal monitors were able to avoid.
The area samples, although not directly indicative of exposures, were useful for corroborating results from the personal samples and providing additional insights regarding exposures. For example, the personal samples suggested that individual particle-phase PAH levels were below limits of detection (0.4–14 µg m–3). This was confirmed by the area samples, which were typically <0.5 µg m–3. Also, the area samples allow us to estimate upper limits for PM2.5 (6.98 mg m–3) and PAH (3.65 µg m–3) exposures during pile burns, using the 8 November data, when the monitor was positioned directly in the smoke plume. Although no regulatory standards currently exist for wood smoke, these concentrations can be compared to the ACGIH (2007) guideline for respirable particulate (3 mg m–3) and the NIOSH (2008) recommended exposure limit for PAHs (100 µg m–3), respectively.
The area samples also allowed us to measure PAH/PM2.5 mass fractions during pile burns, permitting an estimation of PAH exposure for firefighters based on PM2.5 exposure. The mass fractions were nearly twice as high during smoldering (1.06 ± 0.15) than ignition (0.055 ± 0.04 µg mg–1). Fortunately for firefighters, smoldering fires require less attention than flaming fires; hence, firefighter exposures in smoldering fires are substantially less. The higher PAH content in smoldering fires is likely due to the lack of oxygen, which favors incomplete combustion products such as PAHs.
As mentioned previously, PAH/PM2.5 mass fractions can vary with factors such as moisture content, burn type and fuel type. In Fig. 3, we compare our ignition data (µg PAH per mg OC) to ignition data from Lee et al. (2005), who monitored four prescribed burns in Georgia. Both studies took place in predominately pine forests under similar weather conditions. The major difference was the type of prescribed fire: a pile burn in this work versus a broadcast burn in Lee et al. (2005). As shown in Fig. 3, values are within error bars except for the highest molecular weight PAHs (276), which were slightly higher in the Georgia burns. These results suggest that burn type may not be as important as fuel type or moisture content in determining exposure levels.
|
The urinary 1-HP concentrations measured in this study were generally low, and the mean baseline, end-of-shift and next-AM levels were all substantially less than the 90% percentile measurement (0.246 µg l–1) of 1-HP for USA males in 2001 and 2002 (CDC, 2005). Within this context, the non-significant elevation in the baseline urinary 1-HP concentration as compared with the next-AM post-exposure measurement is likely not toxicologically relevant, and may be affected by dietary factors not fully accounted for by questionnaire data. The benefit of using the urinary 1-HP measurement to assess for occupational PAH exposure is that it also takes into account dermal absorption of PAHs, which was not directly monitored in this study. The use of two post-exposure urinary 1-HP measurements in this study, at the end-of-shift and the following morning, allowed for the monitoring of PAH exposure from inhalation and ingestion, with the maximal elimination <8 h following exposure, and dermal exposure, with the maximal elimination 10–15 h following dermal exposure (Buckley and Lioy, 1992; Viau and Vyskocil, 1995).
Personal exposures of wildland firefighters to smoke have been reported previously in the Pacific Northwest (Reinhardt et al., 2000; Slaughter et al., 2004) and California (Letts et al., 1991; Materna et al., 1992; Reh et al., 1994), where assessments focused on carbon monoxide, respirable particulates, aldehydes and acrolein. Materna et al. (1992) also evaluated firefighter exposures to 12 particle-phase PAHs and reported total PAH concentrations between 0.089 and 1.13 µg m–3, values near our PAH detection limits and consistent with our non-detection. Reh et al. (1994) reported not only higher exposures (11.6–35.9 µg m–3) but also included gas-phase PAHs such as naphthalene in their study. Neither study focused on pile burns, and both studies monitored different fuels than in this Southwest US study, lending more evidence that fuel type can affect PAH levels in wood smoke.
The most common non-cancer health effect studies of firefighters have focused on changes in pulmonary function. Although significant declines in lung function have been noted, most studies have indicated transitional lung function effects that disappear after an extended period of little to no occupational smoke exposures. Betchley et al. (1997) reported significant declines in cross-shift and cross-season FVC and FEV1 for Oregon/Washington wildland firefighters, without correlating exposure data. Their work also reported that FEV1 and FEV25–75% recovered more slowly when home fuels included wood. Slaughter et al. (2004) identified significant decrements in lung function (FEV1, FVC and FEF25–75%) for firefighters during prescribed burns in the Northwest (n = 65), but they failed to find an association between any lung function indicator and PM3.5, formaldehyde, acrolein or carbon monoxide.
There are a number of limitations to the current study. Only pile burns were evaluated, so it was not possible to measure PAH exposure during either broadcast prescribed burns or actual wildfires. The relatively small mass of personal particulate collected limited the ability to assess PAH exposure, and although the area monitoring provided a lower detection limit, it did not measure the more volatile PAHs. The measurement of total particulate is consistent with older studies of wildland firefighters; however, more recent studies have used inhalable samplers (Noto et al., 1996). Inhalable samplers would have been useful in this study, too, but we chose the cassette samplers for their ease of deployment and consistency with the PAH samplers. The use of the Institute of Occupational Medicine (IOM) inhalable sampler in this study could have resulted in the measurement of higher dust, but it is unclear whether the larger dust would have had the same particulate PAH concentrations. In addition to PAHs, wood smoke contains a multitude of other chemicals including but not limited to free radicals, aldehydes and metals, so it cannot be stated with certainty that the extent of smoke exposure measured will not result in any adverse health consequences. Because the baseline biological assessments were conducted a minimum of 5 days after any previous firefighting (based on achieving a suitable wood smoke-free period for urinary 1-HP analysis), there may have been some residual effects of previous fires on the initial pulmonary function test, particularly because only 14 of the 21 firefighters had their initial tests prior to any firefighting activities for the season.
In conclusion, exposures to particulate and PAHs did not exceed relevant occupational standards during pile burning in the US Southwest, and no significant changes in urinary PAH metabolite or spirometry were found. Additional studies are needed to better characterize the effects of fuel type and moisture content on exposure levels. Although smoke is a complex mixture and it was not possible to measure all of its components or potential health effects, we were not able to demonstrate excessive or unhealthy exposures on the WMAT reservation, where ponderosa pine is the predominant fuel source for pile burning.
|
FUNDING |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
Native American Cancer Research Project; National Institutes of Health/National Cancer Institute (#5 U54 CA 096281-04).
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
The authors gratefully acknowledge the firefighters of the WMAT who helped us plan for and participated in this study, especially Robert LaCapa, Acting Superintendent, Fort Apache Agency, David Raney (burn boss) and John Pacheco (warehouse and safety supervisor). We thank, the White Mountain Apache Tribal Council, the White Mountain Apache Health Advisory Board and the US Department of the Interior Bureau of Indian Affairs for their assistance. We also thank Marie Hanna-Chatlin, Janlouise Armstrong and Kenna Tessay for their help with sample collection and Glenn Talaska for urinary 1-HP analyses.
Received August 31, 2007; in final form May 1, 2008
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
ACGIH. Documentation of the TLVs and BEIs. American conference of governmental industrial hygienists (2007) Cincinnati, OH.
ATS. Standardization of spirometry: 1994 update. Am J Respir Crit Care Med (1995) 152:1107–36.[Web of Science][Medline]
Aronson KJ, Tomlinson GA, Smith L. Mortality among fire fighters in metropolitan Toronto. Am J Ind Med (1994) 26:89–101.[Web of Science][Medline]
Betchley C, Koenig JQ, vanBelle G, et al. Pulmonary function and respiratory symptoms in forest firefighters. Am J Ind Med (1997) 31:503–9.[CrossRef][Web of Science][Medline]
bin Abas MR, Simoneit BRT, Elias V, et al. Composition of higher molecular weight organic matter in smoke aerosol from biomass combustion in Amazonia. Chemosphere (1995) 30:995–1015.
Birch ME, Cary RA. Elemental carbon-based method for monitoring occupational exposures to particulate diesel exhaust. Aerosol Sci Technol (1996) 25:221–41.[CrossRef][Web of Science]
Boström C-E, Gerde P, Hanberg A, et al. Cancer risk assessment, indicators, and guidelines for polycyclic aromatic hydrocarbons in the ambient air. Environ Health Perspect (2002) 110:451–88.[Web of Science][Medline]
Brown S, Herckes P, Ashbaugh L, et al. Characterization of organic aerosol present at Big Bend National Park, Texas during the Big Bend regional aerosol and visibility observational (BRAVO) study. Atmos Environ (2002) 36:5807–18.
Buckley TJ, Lioy PJ. An examination of the time course from human dietary exposure to polycyclic aromatic hydrocarbons to urinary elimination of 1-hydroxypyrene. Br J Ind Med (1992) 49:113–24.[Web of Science][Medline]
Burnett CA, Halperin WE, Lalich NR, et al. Mortality among fire fighters: a 27 state survey. Am J Ind Med (1994) 26:831–3.[Web of Science][Medline]
Carmella SG, Le KA, Hecht SS. Improved method for determination of 1-hydroxypyrene in human urine. Cancer Epidemiol Biomarkers Prev (2004) 13:1261–4.
Caux C, O'Brien C, Viau C. Determination of firefighter exposure to polycyclic aromatic hydrocarbons and benzene during fire fighting using measurement of biological indicators. Appl Occup Environ Hyg (2002) 17:379–86.[CrossRef][Medline]
Centers for Disease Control and Prevention (CDC). Third national report on human exposure to environmental chemicals (2005) Atlanta, GA: Department of Health and Human Services, National Center for Environmental Health, Division of Laboratory Sciences. (NCEH Pub. No. 05–0570).
Delahunt B, Bethwaite PB, Nacey JN. Occupational risk for renal carcinoma: A case-control study based on the New Zealand cancer registry. Brit J Urol (1995) 75:578–82.[Web of Science][Medline]
Demers PA, Heyer NJ, Rosenstock L. Mortality among firefighters from three northwestern United States cities. Br J Ind Med (1992) 49:664–70.[Web of Science][Medline]
Dockery DW, Pope CA III, Xu X, et al. An association between air pollution and mortality in six U.S. cities. N Eng J Med (1993) 329:1753–9.
Engling G, Herckes P, Kreidenweis SM, et al. Composition of the fine organic aerosol in Yosemite National Park during the 2002 Yosemite aerosol characterization study. Atmos Environ (2006) 40:2959–72.
Ferris BG. Epidemiology standardization project. Am Rev Respir Dis (1978) 108:1–113.
Fine PM, Cass GR, Simoneit BRT. Chemical characterization of fine particle emissions from fireplace combustion of woods grown in the southern United States. Environ Sci Technol (2002) 36:1442–51.[Medline]
Fine PM, Cass GR, Simoneit BRT. Chemical characterization of fine particle emissions from fireplace combustion of woods grown in the midwestern and western United States. Environ Engin Sci (2004) 21:387–409.[CrossRef]
Gaertner RR, Theriault GP. Risk of bladder cancer in foundry workers: a meta-analysis. Occup Environ Med (2002) 59:655–63.
Grimes G, Hirsch D, Borgeson D. Risk of death among Honolulu fire fighters. Hawaii Med J (1991) 50:82–5.[Medline]
Guidotti TL. Mortality of urban firefighters in Alberta, 1927–1987. Am J Ind Med (1993) 23:921–40.[CrossRef][Web of Science][Medline]
Hardin B. Wildfires bring jobs and money to reservations—disasters mean steady work for some Indians. Washington Post, June 16 (2004).
Harner T, Bidleman TF. Measurement of octanol-air partition coefficients for polycyclic aromatic hydrocarbons and polychlorinated naphthalenes. J Chem Eng Data (1998) 43:40–6.[CrossRef]
Hays MD, Geron CD, Linna KJ, et al. Speciation of gas-phase and fine particle emissions from burning of foliar fuels. Environ Sci Technol (2002) 36:2281–95.[Medline]
Hoek G, Dockery DW, Pope A, et al. Association between PM10 and decrements in peak expiratory flow rates in children: Reanalysis of data from five panel studies. Eur Respir J (1998) 11:1307–11.[Abstract]
Kaupp H, McLachlan MS. Gas/particle partitioning of PCDD/Fs, PCBs, PCNs and PAHs. Chemosphere (1999) 38:3411–21.
Kleeman ML, Schauer JJ, Cass GR. Size and composition distribution of fine particulate matter emitted from wood burning, meat charbroiling, and cigarettes. Environ Sci Technol (1999) 33:3516–23.
Lee S, Baumann K, Schauer JJ, et al. Gaseous and particulate emissions from prescribed burning in Georgia. Environ Sci Technol (2005) 39:9049–56.[Medline]
Letts D, Filder AT, Deitchman S, et al. Health hazard evaluation report: HETA 91-152-2140. (1991) Southern California, CA: US Department of the Interior, National Park Service.
Lighty JS, Veranth JM, Sarofim SF. Combustion aerosols: Factors governing their size and composition and implications to human health. J Air Waste Manage Assoc (1995) 50:1565–618.
Materna BL, Jones JR, Sutton PM, et al. Occupational exposures in California wildland firefighting. Am Ind Hyg Assoc J (1992) 53:69–76.[Web of Science][Medline]
Moen BE, Ovebro S. Assessment of exposure to polycyclic aromatic hydrocarbons during firefighting by measurement of urinary 1-hydroxypyrene. J Occup Environ Med (1997) 39:515–9.[CrossRef][Web of Science][Medline]
National Interagency Fire Center. Wildland fire statistics (2008) Available at http://www.nifc.gov/fire_info/fires_acres.htm. Accessed 16 May 2008.
National Wildfire Coordination Group (NWCG). Smoke management guide for prescribed and wildland fire (2001) PMS 420–2, NFES 1279 Available at http://www.nwcg.gov/pms/pubs/SMG/SMG-72.pdf. Accessed 16 May 2008.
NIOSH. Recommended exposure limits for PAHs (2008) Available at http://www.osha.gov/SLTC/coaltarpitchvolatiles/solutions.html. Accessed 16 May 2008.
Noto H, Halgard K, Daae HL, et al. Comparative study of an inhalable and a total dust sampler for personal sampling of dust and polycyclic aromatic hydrocarbons in the gas and particulate phase. Analyst (1996) 121:1191–6.[CrossRef][Medline]
Oros DR, Simoneit BRT. Identification and emission factors of molecular tracers in organic aerosols from biomass burning. Part 1. Temperate climate conifers. Appl Geochem (2001) 16:1513–44.[CrossRef]
Penning TM, Burczynski ME, Hung C-F, et al. Invited review. Dihydrodiol dehydrogenases and polycyclic aromatic hydrocarbon activation: generation of reactive and redox active o-quinones. Chem Res Toxic (1999) 12:1–18.[CrossRef][Web of Science][Medline]
Reh CM, Letts D, Deitchman S. Health hazard evaluation report: HETA 90-0365-2415. (1994) Yosemite National Park, CA: US Department of the Interior, National Park Service.
Reinhardt TE, Ottmar RD, Hanneman AJS. Smoke exposure among firefighters at prescribed burns in the Pacific Northwest (2000) US Department of Agriculture Forest Service, Pacific Northwest Station. Research Paper PNW-RP-526, October 2000.
Ronneberg A, Haldorsen T, Romundstad P, et al. Occupational exposure and cancer incidence among workers from an aluminum smelter in western Norway. Scand J Work Environ Health (1999) 25:207–14.[Web of Science][Medline]
Sama SR, Martin TR, Davis LK, et al. Cancer incidence among Massachusetts firefighters, 1982–1986. Am J Ind Med (1990) 18:47–54.[Web of Science][Medline]
Schwartz J. Air pollution and daily mortality: a review and meta analysis. Environ Res (1994) 64:36–52.[Medline]
Sharkey B. Health hazards of smoke: recommendations of the consensus conference April 1997. (1997) Missoula, MT: US Department of Agriculture Forest Service, Technology & Development Program. (5100 Fire 9751-2836-MTDC). Available at http://www.fs.fed.us/fire/safety/ref_material/content/health_hazards_smoke.pdf. Accessed June 2007.
Simoneit BRT, Rogge WF, Mazurek MA, et al. Lignin pyrolysis products, lignans, and resin acids as specific tracers of plant classes in emissions from biomass combustion. Environ Sci Technol (1993) 27:2533–41.
Simoneit BRT, Rogge WF, Lang Q, et al. Molecular characterization of smoke from campfire burning of pine wood (Pinus elliottii). Chemosphere Global Change Sci (2000) 2:107–22.[CrossRef]
Slaughter JC, Koenig JQ, Reinhardt TE. Association between lung function and exposure among firefighters at prescribed burns. J Occup Environ Hyg (2004) 1:45–9.[CrossRef][Web of Science][Medline]
Tornling G, Gustavsson P, Hogstedt C. Mortality and cancer incidence in Stockholm fire fighters. Am J Ind Med (1994) 25:219–28.[Web of Science][Medline]
Unwin J, Cocker J, Scobbie E, et al. An assessment of occupational exposure to polycyclic aromatic hydrocarbons in the UK. Ann Occup Hyg (2006) 50:395–403.
US EPA. Development of emissions inventory methods for wildland fire, final report (1996) Available at http://www.epa.gov/ttn/chief/ap42/ch13/. Accessed June 2007.
US EPA. Interim air quality policy on wildland and prescribed fires (1998) Available at http://www.epa.gov/ttncaaa1/t1/meta/m27340.html. Accessed June 2007).
Viau C, Vyskocil A. Patterns of 1-hydroxypyrene excretion in volunteers exposed to pyrene by the dermal route. Sci Total Environ (1995) 163:187–90.[CrossRef][Medline]
Wasserfallen JB, Gold K, Schulman KA, et al. Development and validation of a rhinoconjunctivitis and asthma symptom score for use as an outcome measure in clinical trials. J Allergy Clin Immunol (1997) 100:16–22.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||