Driver Exposure to Combustion Particles in the U.S. Trucking Industry

Abstract:

A large study of combustion particle exposures for drivers of diesel-powered trucks was conducted in collaboration with an epidemiologic study of lung cancer outcomes for workers in the trucking industry. Three components of diesel exhaust combustion particles (PM2.5, elemental carbon, and organic carbon) were measured inside the driver cabs of diesel-powered trucks from 36 different trucking terminals across the United States between 2001 and 2005. In-cab particle exposures for drivers assigned to both short and long distance trips were observed, as well as information on the smoking status of the driver, truck characteristics such as age and model, and weather conditions during the sampling session. 

This article summarizes these findings and describes the relationship between exhaust particles and various determinants of exposure. The results suggest that in-cab particle exposures are positively related to smoking, ambient particle concentrations, truck age, and open windows, with other significant modifying factors such as weather. This study represents the largest and most comprehensive exposure assessment of drivers in the trucking industry, encompassing a 4-year period of observations on diesel and exhaust particle exposures nationwide. The results are relevant not only to the occupational group of truck drivers being examined but also to the general population that live, commute, or work within proximity to diesel-fueled traffic or trucking terminals.

METHODS

Driver exposures were measured at 36 truck freight terminals across the United States, sampling for a 1-week period at each terminal and visiting a new terminal approximately every month during 2001-2005. The terminal locations and time periods visited were generated randomly by a computer within predesignated regions of the United States to ensure the final sample was regionally representative of the approximately 150 eligible large trucking terminals (>100 employees) in the epidemiologic cohort. 

There are two general types of truck driving operations at each terminal:local pick up and delivery (P&D) operations, and long haul (LH), which transports trailers to and from terminals several hundred milesaway. Usually, P&D drivers work during the day and LH drivers work at night, although this is not uniformly the case.

Drivers were asked to perform their normal driving activities with a sampling box mounted in the cab to measure particle concentrations in their work environment. All drivers who volunteered for sampling gave informed consent and completed a short questionnaire about their work and smoking activities. The sampling protocol was approved by the Harvard School of Public Health Human Subject Committee.

The particle collectors, their pumps, and a real-time monitor for temperature and humidity (HOBO; Onset Computer Corp, Bourne, Mass.) were mounted in a box housing connected to an external battery and attached to the dashboard on the passenger side of the cab prior to the driver's leaving on his assigned route; it was retrieved as soon as possible after his return. A picture of the monitor setup is provided in Figure 1. The average elapsed time for the integrated particle samples obtained from these boxes was 10.5 hr. Similar methods were used to measure ambient conditions at an upwind site along the perimeter of the trucking terminal from which the driver was dispatched.

To cross-validate measurements, PM2.5 and carbon monoxide (CO) data were collected from monitors in the Environmental Protection Agency (EPA) Aerometric Information Retrieval System (AIRS) network within a 50-100 km radius around each geocoded terminal address. Because of limited temporal coverage in the EPA monitoring network (collected only every other day), the AIRS data could not be matched directly with each individual driver route. For this reason, the radii of monitor values were averaged over the study week and compared with the average driver exposures from each sampling trip.

The LH trucks are generally equipped with air conditioning; however, P&D trucks are not air conditioned and are often driven with the windows open. Unfortunately, direct information was not collected from the driver on whether he had the window open or shut during the observed route; instead, window status based on a combination of in-cab CO2 concentrations and the indoor-outdoor temperature differential were estimated. 

It was hypothesized that the correlation between indoor and outdoor temperatures would be maximized when the window was open, and CO2 levels would be highest when the window was closed. Based on these assumptions, temperature and CO2 for P&D nonsmoking drivers (smoking would have artificially inflated CO2 levels in the cab) were analyzed.

In Figure 2, a clear cutoff point in the distribution of in-cab CO2 levels was distinguished, while the correlation between in-cab and ambient temperature observations was maximized at CO2 levels 700-800 ppm. 

Based on the combined information from the CO2 and temperature analysis, a cutoff point for open-shut windows at 750 ppm was approximated, with values below given the label "open" and values above given the label "shut" Because this analysis was based on data from integrated samplers (session averages and not real-time data) when windows were partially open or closed during a session, a dichotomous variable (open/shut) would not accurately represent the relationship between window status and exposure. Unfortunately, these "partial" cases could not be clearly identified.


PM2.5 was measured by collecting particles on a 37-mm Teflon filter, 0.2 μm diameter pore size, after passing through a precision machined cyclone separator (GK2.05SH [KTL]; BGI, Inc., Waltham, Mass.) to remove particles greater than 2.5 μm aerodynamic diameter. Mass collected on the filter was determined by gravimetric analysis using an analytical balance (Micro-Gravimetric No. M5; Mettler Instruments Corp., Hightstown, N.J.). The filters were preweighted after humidity equilibrium (temperature within 20-23°C and relative humidity 40 ± 5% for at least 48 hr) in a chamber. At the end of sampling, the filter was taken back to the laboratory and reweighed after humidity equilibrium to determine weight gain. This method using the KTL cyclone is equivalent to the EPA PQ200 Federal Reference Method.() EC and OC in PM1 were determined by NIOSH method 5040.

PM1 was collected on a 22-mm quartz tissue filter, preceded by a precision machined cyclone separator (SCC1.062 Triplex; BGI) to remove particles greater than 1.0 μm aerodynamic diameter.

The relationship between EC and mass of diesel emissions varies with operating conditions (other combustion sources also emit EC and OC). Lightly loaded and idling diesel emissions have low EC and high OC and are indistinguishable from general gasoline engine emissions.

However, EC in PM1 was chosen as the primary marker of diesel exposure because of interest in a marker of fresh diesel and other vehicle emissions before they had agglomerated into the accumulative mode of atmospheric aerosols. Measurements made during a feasibility study using side-by-side DustTrak samplers with PM2.5 and PM1 inlet impactors showed that the PM1 mass is normally 90-98% of the PM2.5 mass in the vehicle cabs.

Moreover, EC is not a significant component of cigarette smoke (0.49%) and is a small component of gasoline car engine emissions (2-5 mg/km) and propane emissions when compared with heavy-duty truck engine emissions (early 1990s technology, 164 mg/km).

There was also an artifact associated with OC data where there was cigarette smoking. It is known that quartz fiber filters (used for EC + OC measurements) will adsorb polar hydrocarbon vapors from the airstream and cigarette smoke has a high concentration of polar vapors. In those settings, the OC content frequently exceeded PM2.5.

All the statistical analyses were performed using STATA Version 8.2. The concentration data are not normally distributed, and summary statistics are presented for both the arithmetic and geometric means and standard deviations (Tables I-III). All data comparisons were made using nonparametric statistical tests (Spearman Rho, Wilcoxon rank sum, Kruskal-Wallis) unless otherwise noted. 

The 36 site visits yielded a total of 349 LH and 576 P&D observations for PM1 (EC and OC), and 334 LH and 545 P&D observations for PM2.5. Because PM2.5and PM1 were measured on separate filters, the differences in sample sizes were the result of losses that occurred on only one of the filters during a given session.


There were missing data across the various predictors of in-cab exposures, including substantial missing data on smoking and vehicle characteristics, and no observed data for window status. These missing data complicated the use of multiple regression techniques, and for this reason, the impact of the predictors are estimated and discussed individually. Future site visits will include a revised sampling plan to fill in these data gaps, and the observations on the individual effects will be cross-validated with the newly collected data in a multiple regression context.

RESULTS AND DISCUSSION

Summary statistics of particle concentrations observed in driver cabs are presented in Tables I-III by job description , smoking, and predicted window status, as well as the ambient conditions observed at the home terminal during the sampling trips. Observed levels are low by comparison with a similar study of the industry done in the 1980s(20) that used identical analytical methods and sampling protocol to collect EC and PM2.5 at six large U.S. trucking terminals.

 During the earlier time period, in-cab EC concentrations were much higher (P&D: GM = 4.0, GSD = 2.0; LH GM = 3.8, GSD = 2.3) compared with the levels observed in the present study (P&D: GM = 1.2, GSD = 2.8; LH GM = 1.1, GSD = 2.3). This is likely the result of the downward trend in ambient PM over time, as well as the changing regulatory environment imposing restrictions on fuel composition, idling, vehicle type, etc. The following results describe the individual factors related to driver exposures: smoking, job description, ambient conditions, window status (up or down), weather, and vehicle characteristics.

Smoking

Tables I–III present the summary statistics by smoking status for the two groups of drivers.Smokers were exposed to significantly higher particle concentrations than nonsmokers for bothP&D and LH drivers (p < 0.05 for EC, and p < 0.01 for OC and PM2.5), with the largest effecton OC and PM2.5. The geometric means (Tables I–III) show an approximate doubling of OC and PM2.5 concentrations due to smoking in the truck cabs, while the ratio is much closer tone (no effect) for EC. The latter result is consistent with similar studies showing only a small fraction of EC in cigarette smoke.

Differences by Job Description

Summary statistics by P&D and LH drivers are shown in Tables I–III. There were no significant differences in particle exposures across the two driver groups, with the exception of significantly higher OC and PM2.5 concentrations in the subset of nonsmoking LH drivers compared with nonsmoking P&D drivers (p < 0.01). Although the explanation behind this difference remains unclear, it may be attributable in part to differences in traffic composition between interstate (LH) and intercity (P&D) driving, as well as to increased idling time for LH drivers

Correlation with Ambient Particle Concentrations

Table IV lists the correlation coefficients relating in-cab particle measurements to observed terminal background conditions, as well as EPA AIRS measurements of PM2.5 and CO within a 50–100 km radius of the terminal (not route specific). In-cab particle concentrations were significantly correlated (r = 0.2–0.5; p < 0.01) with background concentrations measured upwind at the driver’s home terminal during that same period. Because P&D drivers remain more local, the association between terminal background and in-cab measurements was stronger for P&D drivers (r = 0.4–0.5; p < 0.01) than for LH drivers (r = 0.2–0.4; p < 0.01).The results also show that the AIRS PM2.5 data is more closely related to in-cab PM2.5 andOC concentrations than it is to in-cab EC levels, while the in-cab EC levels are more highly correlated with the AIRS CO data.

Similar studies support this correlation between EC and CO, pointing to traffic as the common source for both pollutants.() The correlation provides evidence that ambient CO is a better marker of EC than PM2.5, which has important implications for the reconstruction of historical exposures for this cohort. The relatively low correlations observed between in-cab measurements and the EPA monitors suggest that route-specific information is necessary to predict driver exposures more accurately. However, the evidence remains suggestive of a regional component to driver exposures that must be accounted for in a prediction model.

Truck Windows and Exposure


Summary statistics for particles based on the open-shut designation for nonsmoking P&D drivers are provided in Tables I-III. The results provide evidence of an external source for particle exposures, with higher concentrations when the windows are open (p < 0.01 for EC, p < 0.05 for OC and PM2.5). This is consistent with other findings in the literature showing increased particle concentrations from traffic exposures with open windows.

Although the LH in-cab CO2 levels showed a similar distribution and cutoff point, the relationships between a dichotomized open-shut status were not significant (not shown). This is not surprising given the modifying effect of air-conditioning systems in the LH trucks. However, these results are subject to misclassification of exposures, since direct observations were not made on window status. For this reason, future site visits will include a revised sampling plan to collect this information that will be used to cross-validate the findings from the less rigorous method of estimating window status using CO2 and the indoor-outdoor temperature differential.

Wether

Weather data on wind speed, relative humidity, precipitation, temperature, and barometric pressure were observed from the closest monitoring station to the main terminal and matched to the specific sampling time periods using an online source (Weather Underground, Inc., Ann Arbor, Mich.). The weather variables from this source correlated very highly (r > 0.9; p < 0.01) with a limited set of weather observations made by our on-site weather station and were representative of weather conditions at the base terminal during the driver shift. These weather indicators were not significantly related to OC and PM2.5.

However, average wind speed and temperature were both significantly related to in-cab EC concentrations. Increasing temperature, which is likely a surrogate for window status for P&D drivers since there is not a large seasonal component to ambient PM, had a significant and positive effect on EC concentrations (r = 0.1; p < 0.01). Wind speed was negatively correlated with in-cab EC concentrations (r = -0.4; p < 0.01), which is similar to other findings in the literature showing faster winds associated with increased dispersal rates and lower particle concentrations.

Vehicle Characteristics

Vehicle characteristics were obtained on some of the trucks sampled, including production year, make, and model. Although make and model did not have a significant effect on in-cab particle concentrations, EC levels were significantly impacted by truck age (R2 = 0.34; p < 0.01). This relationship is displayed in Figure 3 for the subset of P&D drivers with the median EC value graphed by production year (group medians are more representative of the central tendency due to the non-normality of exposure data). 

This relationship suggests that seepage of diesel exhaust into the truck cab worsens with age (older production years) due to leaks from the cab's rubber seals that allow exhaust from the engine compartment to enter the cab. 

Limitations 

Much of the data on smoking and window status are missing, which complicated the modeling effort. There is also no route-specific information for the majority of the samples, including locations traveled, number of stops, traffic conditions, etc. This information is necessary to match background conditions with the individual driver routes more closely.

For example, a driver will not stay neatly within a given radius of a terminal but will likely exit this radius and drive in one general direction. This has been verified with a limited set of GPS data collected from drivers, suggesting that the use of average location characteristics around the home terminal to predict driver exposures is subject to a high degree of exposure misclassification. For this reason, future sampling efforts will include more GPS tracking units and real-time particle exposure monitors in the vehicle samples, and traffic information will be solicited from drivers immediately following their return to the terminal. This data will allow investigators to match speed, acceleration, elevation, and location with the various exposure peaks and spikes apparent in the real-time data to determine their causes more accurately.
