Occupational Exposure to Diesel Particulate Matter in Municipal Household Waste Workers

ABSTRACT

Objective

The purposes of this study were to determine the following: 1) the exposure levels of municipal household waste (MHW) workers to diesel particulate matter (DPM) using elemental carbon (EC), organic carbon (OC), total carbon (TC), black carbon (BC), and fine particulate matter (PM 2.5) as indicators; 2) the correlations among the indicators; 3) the optimal indicator for DPM; and 4) factors that influence personal exposure to DPM.

Methods

A total of 72 workers in five MHW collection companies were assessed over a period of 7 days from June to September 2014. Respirable EC/OC samples were quantified using the thermal optical transmittance method. BC and PM 2.5 were measured using real-time monitors, an aethalometer and a laser photometer. All results were statistically analyzed for occupational and environmental variables to identify the exposure determinants of DPM.

Results

The geometric mean of EC, OC, TC, BC and PM 2.5 concentrations were 4.8, 39.6, 44.8, 9.1 and 62.0 μg/m3, respectively. EC concentrations were significantly correlated with the concentrations of OC, TC and BC, but not with those of PM 2.5. The exposures of the MHW collectors to EC, OC, and TC were higher than those of the drivers (p<0.05). Workers of trucks meeting Euro 3 emission standard had higher exposures to EC, OC, TC and PM 2.5 than those working on Euro 4 trucks (p<0.05). Multiple regression analysis revealed that the job task, European engine emission standard, and average driving speed were the most influential factors in determining worker exposure.

Conclusions

We assessed MHW workers’ exposure to DPM using parallel sampling of five possible indicators. Of these five indicators, EC was shown to be the most useful indicator of DPM exposure for MHW workers, and the job task, European emission standard, and average driving speed were the main determinants of EC exposure.


MATERIALS AND METHODS

Exposure Group Selection and Task Description

Five Korean MHW collecting companies, three in Goyang and two in Seoul agreed to participate in this study. Goyang is a medium-sized (267.31 km2) suburban city near Seoul with a population of 1 million. Seoul is a metropolitan city with 10 million residents.

In Korea, MHW is classified into three types: solid waste, food waste, and recyclable materials such as plastic, paper, cans, clothes and bottles. All of the companies collect all three types of MHW. Workers who collect recyclable waste were excluded from this study because the recyclable waste trucks use LPG (liquefied petroleum gas). Only MHW workers who use diesel-powered trucks were included in this study. Trucks that collected solid waste went either to their respective incineration plants or to interim collection points such as landfills 2–5 times per day, depending on their route and pick-up locations. The food waste trucks went to their recycling plant several times a day.

A MHW collection truck is manned by 1–2 collectors and a driver. Collectors retrieve the MHW and dump it into the rear compartment of the truck. All of the trucks are equipped with a GPS (Global Positioning System) system, and hydrodynamic presses and have semi-automated systems to lift the trash bins or containers to dump the trash into the trucks. Collectors usually stay in the rear of the truck to dump the trash and to operate the press and lifting mechanisms. All of the exhaust tailpipes of the trash trucks are positioned under and toward the rear of the trucks and the rear of the truck is where workers have the greatest risk of exposure to 3 / 17 Fig 1. Photographs of municipal household waste-collecting activities. Left: Riding on the rear of a truck. Right: Collecting MHW with samplers mounted. DEE (Fig 1). Drivers stayed inside the trucks for more than 6 hours unless they needed to assist the collectors. Drivers would help if there were only one collector.

Sampling Strategy

Field sampling was conducted over a period of 7 days between 26 June and 18 September 2014. The sampling locations, dates, number of samples collected and waste type are listed in S1 Table. Seventy-two EC/OC/TC, 17 BC, and 21 PM 2.5 personal samples were collected from 72 MHW workers. Prior to each sampling date, workers and managerial staff were briefed on the plan, purpose, and method of the sampling, and the majority of the workers agreed to participate. Because of the limited number of available instruments for BC and PM 2.5 sampling, just one to two trucks and their workers were selected for comparative sampling of EC/OC/TC, BC and PM 2.5 during the meeting. To minimize possible sampling bias, we selected the most representative ones after discussing the workload, manning, collection route and locations with the company manager and workers.

On the sampling day, all workers who volunteered for sampling wore an EC/OC/TC sampler. The workers who had previously been selected for comparative sampling additionally wore BC and PM 2.5 samplers, as shown in Fig 1. The sampling was performed during the entire workday. Work schedules differed among the companies and between the two cities. The workday also varied depending on the route and the amount of MHW collected. Typically, a workday and sampling period ranged from 400 to 500 minutes. Since MHW collection is physically demanding, we were unable to collect repeat samples from the same worker. After the sampling was completed, all workers answered a short questionnaire about their employment history, number of service years and smoking habits.

Sampling and Analysis

All samples were collected in the breathing zone of the collectors and drivers. EC/OC/TC samples were collected on 37-mm diameter, pre-fired quartz filters (Pallflex Tissuquartz 2500QAT-UP, Pall Life sciences, USA) mounted on a personal environmental monitor (PEMs, Cat No 761–203, SKC Inc., USA) using a personal sampling pump (MSA Escort ELF pump, Mine Safety Appliance Co., USA). Pumps were pre- and post-calibrated using a DryCal DC-Lite primary flow meter (DCL-H, Bios International Co., USA). According to the PEM manufacturer’s instructions, the pump flow rate was set at 2 L/min. At this rate, PEM samplers have a 50% cut-off point for particulates with an aerodynamic diameter of 2.5 μm. Field blanks were collected daily at the measurement sites and were handled identically to the personal 4 / 17 samples. All samples were sent for analysis to the laboratory of the Occupational Lung Diseases Institute, Korea Worker’s Compensation and Welfare Service. This is the only laboratory in Korea that analyzes EC/OC/TC samples using NIOSH method 5040. The laboratory participates in the American Industrial Hygiene Association (AIHA) Proficiency Analytical Testing (PAT) program. 1.5 cm2 of the quartz filter was punched out and analyzed using an OCEC carbon aerosol analyzer (Sunset Laboratory Inc., USA). The limit of detection (LOD) was 0.2 μg per cm2 filter for both EC and OC. All sample measurements for this study exceeded the detection limit.

BC was measured using an aethalometer (microAeth model AE51, Magee Scientific, USA). This instrument measures the intensity of light (880 nm wavelength) transmitted through a T60 Teflon coated glass fiber and reports BC concentrations in ng/m3. The default manufacture’s specific attenuation coefficient of 16.6 m2/g was used. The air sampling rate was set at 0.15 L/min to enhance the sensitivity per the manufacturer’s manual. Real-time measurements were recorded every minute.

The PM 2.5 concentrations were measured using a real-time laser photometer (SidePak Model AM510, TSI Inc., USA). The SidePak has a built-in PM 2.5 μm impactor. The instrument was set to an airflow rate of 1.7 L/min. All SidePaks used had been calibrated by the manufacturer within the recommended one year interval. Real-time readings were collected every minute. The measured PM levels were corrected using the gravimetric calibration factor, which was determined by collecting parallel samples on PVC filters (37-mm, pore size 5.0 μm, SKC, Inc., USA) mounted on the PEM samplers. Detailed experimental procedures for the determination of the calibration factor are presented in the S1 File.

Ambient Background Levels

Ambient concentrations of EC, OC, TC, BC and PM 2.5 were obtained from the air pollution monitoring stations in Goyang and Seoul and were taken from the database of Air Quality Information of Seoul metropolitan area and GyeongGi-Do. Monitoring stations are located on the roofs of 3‒4 story buildings in residential areas and near main streets. The monitoring station data used in our study were located where the MHW workers made their collections. However, if there was no monitoring station near the collection site, then the data from the closest station were used for the background values.

The Air Quality Information monitors use a semi-continuous OCEC field instrument (Sunset Laboratory Inc., USA) for EC/OC/TC, and aethalometer (model AE22, Magee Scientific Company, USA) for BC. PM 2.5 concentrations were measured by a ß-ray absorption method using a continuous particulate analyzer (SPM 613-D, Kimoto, Japan). All measurements were collected at hourly intervals and the mean concentrations were calculated from the sampling period.

Statistical Analysis

Probability plots of EC/OC/TC, BC and PM 2.5 data were right-skewed and a Kolmogorov-Smirnov analysis of the data indicated that the measurements would be best described by a lognormal distribution. All time-weighted average (TWA) data were natural-log-transformed for statistical analysis, and the geometric mean and geometric standard deviation were used for the mean and standard deviation in the descriptive statistics. Although real-time measurements were made for the PM 2.5 and BC monitors, only TWA values were used in this study. The real-time measurements will be described in later article. The descriptive statistics (geometric mean, geometric standard deviation, minimum and maximum) were calculated. A Pearson’s 5 / 17 correlation analysis was performed to assess the relationships among the log-transformed concentrations of each DPM indicator.

All EC/OC/TC, BC and PM 2.5 results were classified using environmental and occupational variables such as job task (collector vs. driver), waste type (solid vs. food), age of the diesel vehicle (1‒5 yrs, 6‒10 yrs and 11‒15 yrs), diesel engine emission standard (Euro 3 vs. Euro 4), truck payload capacity ( 2.5 ton vs. 5 ton), diesel particulate filter (DPF) (factory-installed vs. retrofitted), location (suburban vs. urban), number of collected truck containers (1‒2 vs. 3‒ 4) and worker smoking habits (smoker vs. non-smoker). An analysis of variance (ANOVA) and t-test were used to evaluate the variability within and between the categories of occupational and environmental variables and to compare average levels among categories of occupational and environmental variables.

Multiple regression analysis was performed to identify the main exposure determinants for EC and OC. Categorical variables with p-value <0.05 in the ANOVA were included in a multiple regression analysis. In addition, continuous variables were investigated using univariate analysis, and significant variables with p-value<0.05 entered into a multiple regression analysis. The categorical variables analyzed were job task (collector vs. driver), waste type (solid vs. food), diesel engine emission standard (Euro 3 vs. Euro 4), DPF (factory-installed vs. retrofitted), smoking habits (smoker vs. non-smoker), city (Goyang vs. Seoul), and location (suburban vs. urban). The continuous variables analyzed were driving distance (km), average driving speed (km/h), and percentage of slow driving (< 20 km/h) during the sampling period, weight of collected waste (tons), truck age (y), and truck payload capacity (tons). A multiple linear regression model with the backward elimination method was used. For the final models, differences were considered significant at p<0.05. Model diagnostics were performed with plots of residuals against predicted values and using standardized normal probability plots. Statistics analysis was performed using SPSS 20.0 software (IBM, Armonk, NY).

RESULTS

A total of 72 EC/OC/TC, 17 BC and 21 PM 2.5 measurements were made during MHW collections of solid and food waste. Table 1 shows the TWA values for EC, OC, TC, BC and PM 2.5 for each company. The TWA values for each worker are presented in S2 File. None of the EC and OC measurements were below substance analytical LODs. All measurements were higher than the ambient background levels. The average ratio of exposure level to background level for EC, OC, TC, BC and PM 2.5 was 4.1, 12.7, 9.8, 2.0 and 4.4, respectively. Ambient background levels are shown in Table 1 and the background levels for each day of sampling are listed in S2 Table.

Filter samples of EC TWAs ranged from 1.7 to 29.0 μg/m3 with a geometric mean of 4.8 μg/m3 and the OC TWAs ranged from 13.5 to 107.8 μg/m3 with a mean of 39.6 μg/m3. Real-time measurements for BC had TWAs that ranged from 6.0 to 19.6 μg/m3 with a mean of 9.1 μg/m3. The real-time measurement TWAs for PM 2.5 ranged from 27 to 240 μg/m3 with a mean of 62 μg/m3. T-test results showed that the OC and PM 2.5 levels were significantly different between Goyang and Seoul (p<0.05), but the EC and BC levels were not significantly different. 

Relationships between DPM concentrations and various exposure factors

Table 2 presents a comparison of the EC, OC, TC, BC and PM 2.5 concentrations among occupational and environmental categories. The mean EC (N = 42, 5.6 μg/m3), OC (44.2 μg/m3), and TC (50.1 μg/m3) for MHW collectors were significantly higher than those for drivers (EC, N = 30, 3.8 μg/m3, p = 0.003; OC, 34.1 μg/m3, p = 0.015; TC, 38.3 μg/m3, p = 0.008). This indicates that the job task significantly influenced personal exposure levels of EC, OC and TC. Similarly, the mean BC (N = 10, 10.1 μg/m3) and PM 2.5 (N = 11, 68.6 μg/m3) for the collectors were slightly higher than those of the drivers (BC, N = 7, 7.8 μg/m3 and PM 2.5, N = 10, 55.6 μg/m3), albeit not significantly so.

All MHW trucks surveyed had manufacture dates after 2000. Their average age was 8.2 y and they met either the Euro 3 or 4 diesel engine emission control standards. The ANOVA analysis results indicated that the engine emission standard (Euro 3 vs. Euro 4) was a significant factor affecting personal exposure levels to EC, OC, TC and PM 2.5, whereas the age group of the truck was not. The workers using Euro 3 Standard trucks were exposed to significantly higher levels of EC (N = 41, 5.6 μg/m3, p = 0.004), OC (45.0 μg/m3, p = 0.005), TC (50.8 μg/m3, p = 0.004) and PM 2.5 (N = 6, 96.2 μg/m3, p = 0.037) than those working on the Euro 4 trucks (EC, N = 31, 3.9 μg/m3; OC, 33.5 μg/m3; TC, 38.0 μg/m3; PM 2.5, N = 15, 52.1 μg/m3). Those working on trucks with a payload capacity equal to 5 tons had significantly higher exposures to OC (29.9 vs. 43.0 μg/m3, p = 0.004), TC (35.7 vs. 47.8 μg/m3, p = 0.016) and PM 2.5 (40.9 vs. 84.8 μg/m3, p = 0.004) than those on trucks with a payload capacity of less than 2.5 tons. No such relationship was found for the EC and BC data. 7

All of the exhaust tailpipes of the MHW trucks were positioned under and toward the rear of the trucks. The distance from the tailpipe to the rear of the truck varied from 1.2 to 4.2 m, depending on the truck model. The newer trucks had greater distances between the tailpipe and the rear of the truck. The collectors working on trucks with greater distances ( 4 m) between the tailpipe and the rear of the truck had lower EC exposures than the collectors who 3 worked on trucks that had tailpipes closer than 4m to the rear of the truck (6.3 vs. 3.6 μg/m , p = 0.010). However, this relationship was not observed for OC, TC, BC and PM 2.5 measurements.

The number and quantity of waste of the collections was a significant factor for the DPM exposure levels. Workers who collected more containers (3‒4 vs. 1‒2) had significantly higher exposure levels to EC (5.8 vs. 4.1 μg/m3, p = 0.007), but there was no significant difference for OC, TC, BC and PM 2.5. The workers who smoked during the sampling period had mean exposures to OC (49.3 vs. 30.6 μg/m3, p<0.001) and TC (54.5 vs. 35.4 μg/m3, p<0.001) that were significantly higher than those of the non-smokers, but there was no significant difference in their exposures to EC, BC and PM 2.5.

Fig 2 shows plots of mean TC levels for job tasks, smoking habits and vehicle factors. The mean levels of TC for the collectors, smokers, workers on larger trucks, and on trucks meeting Euro Standard 3 were significantly higher than the levels of the drivers, non-smokers, workers on smaller trucks, and those working on trucks meeting Euro Standard 4. Fig 2 also shows the ratio of OC to EC at the end of each column, which ranged from 1.4 to 26.1, with a mean of 8.2. The mean ratio of OC to EC for smokers, workers on larger trucks, and workers on trucks that had greater distances between the tailpipe and rear of the truck was significantly higher than those for the other categories of workers. This indicates that the former workers were exposed to significantly higher fractions of OC compared to EC.


Correlations between DPM indicators

The concentrations of EC were significantly correlated with the concentrations of OC, TC and BC, indicating a consistent pattern among representative DPM indicators (Table 3). The Pearson correlation coefficients between EC levels and OC, TC, and BC were 0.325 (p<0.01), 0.468 (p<0.001), and 0.822 (p<0.001), respectively. PM 2.5 levels showed significant correlations with OC and TC, but not with EC and BC. Since TC is the sum of EC and OC, the significant correlation between PM 2.5 and TC is also related to the correlation between OC and PM 2.5.

Multiple linear regression analysis

Table 4 summarizes the results of the multiple linear regression analysis performed to identify exposure determinants affecting the levels of EC and OC. The EC multiple regression model included seven variables related to the vehicle, worker activity, and environment. The factors included in the multiple regression analysis were selected after performing a univariate analysis using a significance level of 0.05. The univariate analysis results were: job task (ß = 0.387, p = 0.003), Euro engine emission standard (ß = -0.376, p = 0.004), truck age (ß = 0.043, p = 0.024), number of truck containers collected (ß = 0.267, p = 0.008), percentage of slow driving (< 20 km/h) during the sampling period (ß = 2.146, p = 0.014), average driving speed (ß = -0.038, p = 0.024), and the ambient background level (ß = 0.043, p = 0.682). The background level was applied to adjust for the effects of ambient levels. Ambient values vary depending on the amount of traffic and the occasional Asian dust event [21, 22]. The variables were selected based on the backward elimination method for the multiple regression model. The final model to predict EC exposure levels included job task, Euro engine standard, and average driving speed (adjusted R2 = 0.382, p<0.001).

Six variables were included in the OC model; job task (ß = 0.261, p = 0.015), Euro engine emission standard (ß = -0.295, p = 0.005), truck payload capacity (ß = 0.140, p = 0.004), smoking (ß = 0.094, p<0.001), city (ß = -0.397, p = 0.003), and ambient background level (ß = -0.063, p = 0.198). The final model to predict the OC exposure level included smoking, Euro engine standard, job task and truck payload capacity (adjusted R2 = 0.470, p<0.001).

