Applied Occupational and Environmental Hygiene

Abstract:

The purpose of this study was to characterize respirable dust, crystalline silica, diesel, and noise exposures to construction workers on a large highway construction project in Boston, Massachusetts. The study primarily focused on operating engineers and laborers, and to a lesser extent on ironworkers and carpenters during the tunnel finish and cut and cover stages. Full and partial shift noise dosimeter measurements were collected. Eighty percent of the 40 noise measurements were at or above 85 dBA, with the operating engineers averaging 91 dBA.

Sample collection for respirable dust, crystalline silica, and diesel particulates measured as elemental carbon was done using cyclones and personal cascade impactors. Exposure to respirable dust ranged from 0.06 to 21.77 mg/m3 for the 77 workers sampled, with the laborers having the highest mean concentration of 2.85 mg/m3 .

The respirable quartz measurements for the 32 samples colected ranged from the detection limit of 0.008 mg/m3 to 1.64 mg/m3, with the highest mean concentration of 0.31 mg/m3 attributed to the laborers. The use of drills, when compared to other machine types, produced the highest exposures to respirable quartz. The levels for elemental carbon ranged from 4 to 178 g/m3 (mean of 41 g/m3) inside enclosed work spaces, compared with 0.5 to 53 g/m3 (mean of 10   g/m3) for samples collected in unenclosed work sites.

Statistical modeling of the factors influencing the elemental carbon exposures showed that machine type, worker distance from the diesel source, the number of other diesel sources in the work area, and site enclosure were all significant exposure determinants. The results indicate that high exposures to noise, respirable dust, and crystalline silica are common in the highway construction industry.

MATERIALS AND METHODS

Noise

Thirty-six personal noise dosimetry measurements were collected among different individual workers using Quest Q-100 noise dosimeters (Quest Electronics, Oconomowac, Wisconsin), with an 80 dBA threshold level. Half of the samples were collected for approximately one hour, while the remaining noise measurements were for six-hour intervals. Extrapolation to an 8-hour time-weighted average (TWA) was performed assuming that the noise level measured during the sample period would be consistent throughout the entire work shift.


This assumption was considered valid because workers often performed similar tasks throughout an entire workday, a trend observed by the industrial hygienists on site. The noise measurements represented several trades in a variety of tasks, including laborers performing drilling and chipping activities, as well as operating engineers running heavy equipment. 

Area samples were collected to evaluate the general noise levels for the large number of tile setters and other trades people completing work within enclosed areas. Each noise dosimeter was calibrated before and after sampling using a Quest QC-10 calibrator. The daily integrated noise measurements were downloaded to a printer, producing a time history of noise exposure. Although specific tasks and areas were chosen for measurement, isolation of contributing noise sources was not possible.

A questionnaire was administered to a subset of the workers (n = 26) wearing the noise dosimetry equipment to determine their attitudes concerning hearing protection devices (HPDs) and hearing loss. Various unions were represented in this sample including operating engineers, laborers, pile drivers, ironworkers, carpenters, and tunnel workers. 

Questions referred to the workers' perception and usage of HPDs. The following were questions of particular interest: How often do you wear hearing protection? Do you believe that if you continue to work in this environment you will develop some form of hearing loss? Do you believe that wearing a hearing protection device will reduce your long-term hearing loss? Answers were multiple choice, representing a range of responses.

Particle Collection

Personal and area particulate samples were collected during field visits to the ITT, cut and cover, and outdoor areas. A number of variables with the potential to modify exposures were collected concurrently with the particulate samples. These variables were recorded through the observations of the on-site personnel collecting the samples. A small percentage of the personal samples were repeat samples collected on different days for the same individual.

Respirable dust samples were collected on either tared 5 mm PVC filters (MSA, Pittsburgh, Pennsylvania) or tared quartz filters (Pallflex Products Co., Putnam, Connecticut) and housed in either two- or three-piece plastic cassettes.

Sample collection was performed using either a BGI-4 Higgins & Dewell (Waltham, Massachusetts), an MSA 10 mm nylon Dorr-Oliver (Pittsburgh, Pennsylvania), or an SKC Aluminum (Eighty Four, Pennsylvania) personal cyclone at a sampling flow rate of 2.2, 1.7, or 1.9 liters per minute (lpm), respectively. 

Particle size selective samples were collected using the Sierra 290 Series personal cascade impactor (Graseby Anderson, Inc., Smyrna, Georgia). The upper stages were collected on mylar substrate sprayed with 316 Silicone Release Spray (Dow Corning Corporation, Midland, Michigan) to prevent particle bounce.

The final stage media was either a 5 mm PVC filter or a quartz filter. The impactor was set up with either the 5A stage, or the 3 and 5A stages, and the backup filter in place. At a flow rate of 2 lpm, the5A stage has a cutpoint of9.8lm and the 3stage has a cutpoint of 3.5 l m, each with a geometric standard deviation (GSD) of 1.21

Because the backup filter of the impactor primarily collects particles less than 3.5 l m, it is an under- estimate of the respirable size selective mass fraction defined by ACGIH. 

Nevertheless, when the data were examined by union or machine type, an analysis of variance procedure found no significant difference in the logarithm of the dust concentration of the impactor backup filter and the cyclone (Tables II and III).

Therefore, for the purposes of this article, the particulate samples collected with the cyclone and the impactor backup filter are combined to provide an estimate of the respirable particulate exposures of construction workers.

To evaluate whether diesel exhaust fume was agglomerating or adsorbing onto larger construction particulate, six inhalable particulate samplers were paired with respirable (cyclone) samplers to collect side-by-side diesel exhaust samples. The inhalable sampler was an adaptation of the UK 7 hole sampler with the particulate collected on a 37 mm quartz filter at a sampling flow rate of 2 lpm.

Air flow pumps for all samples were calibrated indoors prior to and following site visits using a precision rotometer. When sampling was conducted at a substantially different temperature than that at the time of calibration, an adjustment to the measured air volume was made to obtain the actual air volume sampled.

Sample Analysis

Gravimetric analysis of the respirable, inhalable, and particle size selective samples was initially conducted at the Wisconsin Occupational Health Laboratory (WOHL) in Madison, Wisconsin, and, subsequently, at the University of Massachusetts Lowell Industrial Hygiene Laboratory using NIOSH Method 0600.

Following gravimetric measurement, the 5 mm PVC filter samples were analyzed for crystalline silica. Determination of the crystalline silica content was conducted primarily at the WOHL using x-ray diffraction (NIOSH method 7500) with only a few samples analyzed by Liberty Mutual Analytical Laboratory (Hopkinton, Massachusetts) using infrared absorption spectroscopy (NIOSH Method 7602)
  
Elemental carbon analysis was used as a marker of diesel exhaust to avoid contamination by cigarette smoke or other combustion or construction sources. A similar approach was used in a study of diesel exhaust exposures in the trucking industry.

The samples collected on quartz filters were analyzed for elemental and organic carbon by Sunset Laboratory (Forest Grove, Oregon) utilizing a thermal-optical technique. The filters were refrigerated after sampling and shipped to the laboratory on ice.

The thermal-optical analytical technique involves a two- step procedure. The samples are put into an oxygen-free helium atmosphere and heated in four increasing temperature steps to remove all organic carbon from the filters.

A laser is used to monitor the accidental pyrolytic conversion of organic carbon to elemental carbon. The organic compounds are oxidized to carbon dioxide and detected by a flame ionization detector (FID). 

After cooling, a two percent oxygen/helium mixture is switched into the sample oven and the oven is stepped up to 850±C for analysis of elemental carbon, again by FID.

Most of the samples were well above the limit of detection of 0.2 l g/m3 for elemental carbon.

Statistical Analysis

The sampling results and field data variables were entered into a FOXPro database (Microsoft Corporation, Redmond, Washington) and then downloaded into SAS software for the PC (Statistical Analysis System, Cary, North Carolina).

Contaminant analyses reported by the laboratory as less than the detection limit were included in the statistical analyses by assigning the sample the value of the detection limit divided by the square root of two.

There were two elemental carbon and four silica samples at the limit of detection. A total of 68 elemental carbon, 102 respirable dust, 51 quartz, and 47 percent silica samples were analyzed with the SAS software.

Examination of the data through log and normal probability plots and the use of the Shapiro-Wilk's statistic showed that the samples fit a lognormal distribution better than a normal distribution. 

Thus, all statistical analysis includes the geometric mean and standard deviation calculated from the logarithmically transformed mass concentration values.

A statistical model was developed to estimate the effect of various exposure modifiers on the natural logarithm of the elemental carbon levels. The modeling effort began with a single factor analysis of variance (ANOVA) of the nine variables suspected of influencing the personal diesel exposure level: the machine type (n = 4), the number of sides enclosed on the machine (0, 3, or 4), the presence/absence of an exhaust scrubber, the amount of time the machine was actually performing work (0 to 1.5 hours, 2 to 3.5 hours, or 4 to 5.75 hours), the degree of enclosure of the site (closed versus open), the distance from the source of the exhaust to the sampling media (< = 10 feet, 10 to 20 feet, or >20 feet), the number of other diesel sources in the area (< = 1 or > 1), the union to which the worker belonged, and whether or not the worker smoked cigarettes during that workday.

Multivariate ANOVA models were then created using a forward selection method, beginning with the variable with the lowest p-value and adding variables until further additions did not result in statistically significant p-values for the added variable or earlier variables lost statistical significance. A second multivariate statistical model was created using a backward elimination methodology. All nine variables were entered into the initial model. One by one, the variables having the highest p- values were removed. Both the forward selection and backward elimination methods produced the same final model for the determinants of elemental carbon exposures.

Once the final model was accepted, the model was rerun as a multivariate linear regression using dummy variables to represent each level of the categorical variables. This approach produces unique parameter estimates for each level of each categorical variable.

RESULTS

Data collection in Boston began after the immersed tunnel tubes (ITT) were in place and the land tunnels were well into the construction process. Thirty-two field visits were conducted at random between June, 1994, and April, 1995.

Exposure assessment activities initially began inside the ITT during the sum- mer months and then were expanded to two other contract sites, allowing the evaluation of outdoor and cut and cover work as well. The winter months (February and March) provided the opportunity to conduct exposure assessment activities within cut and cover areas that were enclosed and heated to allow for the correct curing of the concrete.

Noise

A total of 40 noise dosimetry measurements were taken at the various contractor sites (Table I). Nearly half of the samples (48%) were at or above 90 dBA. Eighty percent of the measurements were at or above 85 dBA. The average noise levels were consistently high among the various trades, ranging from a mean of 86 dBA for carpenters to a mean of 91 dBA for operating engineers.

Of the 26 workers responding to the noise questionnaire, 73 percent acknowledged that HPDs were offered to them on a regular basis. In addition, 89 percent reported that there was a place nearby to get HPDs if they were needed. The workers were then asked to estimate how often they used HPDs on site by choosing one of four categories: never, sometimes, often, or always.

The majority of the workers (69%) reported they "sometimes" wore HPDs. Twelve percent reported "never" using HPDs and 19 percent said they "often" wore HPDs. The infrequent use of HPDs reported by workers occurred despite the perception of 73 percent of them that it was, at least, "very likely" that they would develop some form of hearing loss if they continued to work in their present environment.

Further, 85 percent of the participants believed that it was, at least, "very likely" that HPDs would reduce their long-term hearing loss.  

Dust and Silica

Examination of the 77 respirable dust samples indicated exposure levels ranging from 0.06 to 21.77 mg/m3 (Table II). Twelve percent of the samples exceeded the OSHA PEL of 5 mg/m3 for particulate not otherwise regulated (PNOR) and 13 percent exceeded the ACGIH TLV of 3 mg/m3. 

The laborers had the highest mean respirable dust exposure of 2.85 mg/m3 while performing tasks requiring the use of drills, grinders, and chip- ping guns for concrete removal, as well as the concrete mixer.

Of the 77 workers sampled, 42 percent wore some type of respiratory protection. A protection factor of 5 was assigned to the single-use dust masks and a 10 to the disposable HEPA half masks worn by the trades.

After incorporating the protection factors, only two workers, both of whom were not wearing respirators, had exposures over 3 mg/m3.

The percentage of silica in the respirable samples varied from 27 percent down to the limit of detection, which for these samples was 0.7 percent. On average, the percentage of silica was similar among the various trades. However, because the dust exposure of the laborers was greater, the average silica exposure was correspondingly higher. 

Concrete work, including chipping, drilling, and mixing concrete, was the primary source of silica on the construction sites. Residual dust created during these activities had the potential to remain on site for long periods of time, affecting others not directly involved in concrete work.

The OSHA PEL takes into account the percentage of respirable silica in each sample, reducing the allowable respirable dust PEL accordingly. Fourteen workers (44%), thirteen of whom were laborers, had exposures exceeding the calculated OSHA PEL for respirable-dust-containing silica. After incorporating the respiratory protection factors, the exposures of six workers remained above the PEL.

The 35 respirable quartz measurements ranged from 1.64 mg/m3 down to the limit of detection of 0.008 mg/m3. On average, the laborers had the highest exposures compared to the other trades. Exposures exceeding the NIOSH REL for silica of 0.05 mg/m3 included 63 percent (17) of the laborers, 40 percent (2) of the operating engineers, and 50 percent (1) of the ironworkers.

Similarly, 48 percent (13) of the laborers and 50 percent (1) of the ironworkers had exposures exceeding the ACGIH TLV for quartz of 0.10 mg/m3. Fifteen of the twenty workers with exposures of crystalline silica over 0.05 mg/m3 wore some type of respiratory protection. After incorporating the protection factors offered by the single-use dust masks and disposable HEPA half masks worn at the time of sample collection, 29 percent (10) of the workers still had exposures exceeding 0.05 mg/m3 and 17 percent (6) remained above 0.10 mg/m3.

The equipment used to perform a given task can greatly affect the resultant exposure level (Table III). Each of the machine types listed in the table was used in conjunction with concrete work, however, the exposures created varied widely. Drilling produced the highest geometric mean for respirable quartz at 0.43 mg/m3 , with 80 percent of the samples exceeding the OSHA PEL for respirable dust and 100 percent exceeding 0.05 mg/m3 of crystalline silica. 

Tasks performed with a chipping gun led to the second highest mean quartz exposures, with 40 percent of the samples exceeding the OSHA PEL and 50 percent exceeding 0.05 mg/m3 of crystalline silica. The mean silica percentages ranged from 3.4 percent to 16.2 percent among the various ma- chine types, with the drill averaging 12.0 percent.

Exposures exceeding the current silica standards were not ob- served among workers doing concrete removal with large fork-lifts having grinder or scabbler attachments, most likely due to the use of hoses to wet down the concrete and the greater distance from the source of silica dust to the worker. With the exception of core drilling, for which two samples were collected, all of the drilling and chipping tasks were performed on dry concrete without the aid of wet methods.

The respirable dust and silica concentrations for the personal samples collected during concrete mixing were below occupational exposure limits despite the visible dust cloud surrounding the laborer performing this task. Two cascade impactor samples were collected concurrently during a concrete mixing operation and the results compared. One was a personal sample collected the breathing zone of the laborer mixing the concrete. 

A portable fan was set up behind the worker to blow the visible dust away from his breathing zone as bags of concrete were picked up and tossed onto the mixer screen where they were broken open and dumped. An area sample was also collected eight feet downwind of the mixer in the path of the portable fan.

The overall respirable dust concentration of the personal sample was six times higher than the area sample (6.13 mg/m3 versus 0.91 mg/m3). However, for the personal sample, only 13 percent of the total mass was in the respirable range(<3.5l m), and for the area sample 30 percent was in the respirable range. Thus, the fan may have had an impact on reducing the operator's exposure to respirable sized particles. 

Diesel

A total of 68 diesel samples were collected and analyzed for elemental carbon (Table IV). Samples taken during work performed inside the enclosed areas were significantly higher than those taken outside in the open air (paired t-test, T < 0.0005, p = 0.05).

This is evident even after the samples are stratified by trade. The laborers' geometric mean elemental carbon level of 37 l g/m3 for enclosed areas was high relative to that of the operating engineers (26 l g/m3) and the ironworkers (24 l g/m3). 

The geometric means for all of the trades working outdoors were similarly low, ranging from 2 to 9 l g/m3. The highest exposure group (43 l g/m3) consisted of two carpenters working indoors, one with an exposure of 178 l g/m3 while working within five feet of a forklift.

Although isolated diesel particulate is known to be well within the respirable size fraction (MMD = 0.05 l m to 0.3 l m),(13) a subset of six side-by-side respirable- and inhalable-size selective samples were collected to evaluate the potential for diesel particles to agglomerate or to adsorb onto larger construction particulate. These samples ranged from 2 to 55 l g/m3 for general particulate level. 

In four of the five sample pairs, the respirable and inhalable samples revealed similar concentrations with differences of only 0 to 2 l g/m3. For one sample of a diesel platform lift, the respirable result was actually higher by 31 l g/m3 than the inhalable sampler. A paired t-test of the two sampling methods suggested no statistically significant difference (T = 0.34, P = 0.05).

Statistical Modeling of Diesel Exposure Determinants

Statistical modeling of the determinants of elemental carbon level found the type of machine, the distance from the diesel source to the sampling media, the number of other diesel sources in the area, and the enclosure of the work site, to be significant predictors of diesel exhaust exposure (Table V).

The model predicts that the elemental carbon concentration will decrease as the distance from the worker to the source of diesel exhaust increases. Two or more additional diesel sources in the area will elevate the concentration of elemental carbon, as will enclosing the work site. The model explains 66 percent of the variability among the elemental carbon exposures.

By multiplying the parameter estimate by a value of one for the factors which are present and zero for those that do not apply, an estimate of the natural logarithm of the elemental carbon concentration can be calculated for any combination of exposure determinants. To investigate the impact of the exposure modifying factors, a "baseline" working condition was defined and only the category levels for a single category were allowed to vary. Parameter estimates for category levels representing a baseline working condition were multiplied by one and the remaining levels in that category were multiplied by zero.

This was then combined with the parameter estimate for the category level targeted for examination. In this way, the effect of altering the target factor on the estimated exposure concentration can be examined.

The baseline working condition used in the first example (Figure 1) included the following values for the factors in the model: machine type= lift and degree of site enclosure= closed. 

To estimate changes to the elemental carbon concentration, the values for the number of other sources (0 or 1, more than one) and for the distance from the source to the sampling media (<= 10 feet, >10 feet and <= 20 feet, >20 feet) were entered into the model. For example, the model corresponding to a distance of less than or equal to 10 feet and more than one other source resulted in the following equation:

The sum of this equation is 4.28, exponentially this results in an estimated elemental carbon concentration of 72 l g/m3. In the second example (Figure 2), the baseline working condition was changed by holding the distance constant at less than or equal to 10 feet and the number of additional sources at < = 1. The model was then evaluated for each machine type and degree of site enclosure under the baseline working condition by combining the parameter estimates for the baseline with the parameter estimates for the specific machine type and degree of site enclosure.

It is evident from Figure 2 that the degree of site enclosure greatly affects the resultant elemental carbon concentration. Diesel exposure inside an enclosure is over five  times higher than a similar working condition in the outdoors as estimated by the above model.

Estimates of exposure modifiers were determined by the observations of the industrial hygienist conducting the sampling. However, the movement of machinery about a work site creates difficulties in properly and accurately recording the distance from the source of diesel exhaust to the sampling media.

A more accurate characterization of exposure could be obtained by using a work sampling method which more precisely records a worker's distance from the source of exhaust. This is also true of other variables such as wind direction, idle and load time of equipment, and number of other sources, which are subject to change throughout the workday, making a simple measure difficult to quantify.

An adaptaion of the work sampling method, the Time Variant Exposure Analysis (TVEA) method is being used to collect this information in a more systematic way.TVEA is a work sampling method designed to collect information on exposure determinants that vary across the day by recording this data at set time intervals throughout the sampling period. This data is then used to develop estimates of the portion of the sampling period each potential exposure determinant was operational. 
