Determinants of respirable crystalline silica exposure among stoneworkers involved in stone restoration work

ABSTRACT

Objectives:

Crystalline silica occurs as a significant component of many traditional materials used in restoration stonework, and stoneworkers who work with these materials are potentially exposed to stone dust containing respirable crystalline silica (RCS). Exposure to RCS can result in the development of a range of adverse health effects, including silicosis and lung cancer. An understanding of the determinants of RCS exposure is important for selecting appropriate exposure controls and in preventing occupational diseases. The objectives of this study were to quantify the RCS exposure of stoneworkers involved in the restoration and maintenance of heritage properties and to identify the main determinants of RCS exposure among this occupational group.

Methods:

An exposure assessment was carried out over a 3-year period amongst a group of stonemasons and stone cutters involved in the restoration and maintenance of heritage buildings in Ireland. Personal air samples (n = 103) with corresponding contextual information were collected. Exposure data were analysed using mixed-effects modelling to investigate determinants of RCS exposure and their contribution to the individual’s mean exposure. Between-depot, between-worker, and within-worker variance components were also investigated.

Results:

The geometric mean (GM) RCS exposure concentrations for all tasks measured ranged from <0.02 to 0.70mg m−3. GM RCS exposure concentrations for work involving limestone and lime mortar were <0.02–0.01mg m−3, tasks involving granite were 0.01–0.06mg m−3, and tasks involving sandstone were <0.02–0.70mg m−3. Sixty-seven percent of the 8-h time-weighted average (TWA) exposure measurements for tasks involving sandstone exceeded the Scientific Committee on Occupational Exposure Limits recommended occupational exposure limit value of 0.05mg m−3. Highest RCS exposure values were recorded for the tasks of grinding (GM = 0.70mg m−3) and cutting (GM = 0.70mg m−3) sandstone. In the mixed-effects analyses, task was found to be significantly associated with RCS exposure, with the tasks of grinding and cutting resulting in average exposures of between 32 and 70 times the exposures recorded for the task of stone decorating. The between-depot, between-worker, and within-worker variance components were reduced by 46, 89, and 49%, respectively, after including task in the mixed effects model.

Conclusions:

Restoration stoneworkers are regularly overexposed (compared with 0.1 and 0.05mg m−3 8-h TWA) to RCS dust when working with sandstone. The results indicate that the tasks of cutting and grinding sandstone are predictors of increased exposure to RCS dust. In order to decrease exposure to RCS, efforts should be focused on developing and implementing interventions which focus on these high-risk tasks.

MATERIALS AND METHODS

Study population and site description

The study population consisted of 35 restoration stoneworkers employed with the Commissioners of Public Works Ireland, responsible for managing, maintaining, and restoring over 740 of Ireland’s national monuments. Worker participation was voluntary, and subjects were recruited in coordination with the organization’s Health and Safety unit and depot managers.

The stoneworkers were located in six centralized depots and associated historic monuments around Ireland. During this study, workers worked on sandstone, limestone, lime mortar, and granite. The tasks carried out and the materials used in a depot at any one time were dependent on the materials and the type of restoration required for the monument under restoration. Minor restoration of a monument only involved repointing, but major restoration required restoration and/or construction of existing or new sections of the monument. The stoneworkers worked in a stone cutting workshop located in the depot or on site at the monument under restoration depending on the task they were carrying out. Within the group restoration stoneworkers, two main job titles could be distinguished based on their job specifications: stone cutter and stonemason (Table 1). Each depot had an average of one stone cutter and seven stonemasons (range: 1–17).

Walk-through survey

Prior to the commencement of the study, the researcher met with depot managers and workers and presented an overview of the project objectives and proposed sampling protocols. Study participants were asked to complete a self-administered questionnaire, designed using information gathered during preliminary site visits and previous research in this area (HSE, 2001, Lumens and Spee, 2001; Flanagan et al., 2003; HSE, 2006, 2009). The purpose of this questionnaire was to collect information on the tasks, specific tools, materials, and clean-up methods frequently used by the worker in order to develop a contextual information sheet for field sampling. This field sheet was used to collect detailed contextual information during the measurement period including details about the task, tools, materials, exposure controls, respiratory protective equipment (RPE) and other information the author deemed important to record.

At the beginning of the study, a visual assessment was carried out in each of the stone cutting workshops using a checklist to collect detailed information on exposure controls used. This checklist was developed with reference to similar studies (HSE, 2006; Renton et al., 2010) and collected information on the following: design of local exhaust ventilation (LEV) system, maintenance and effectiveness of the LEV including capture velocity, worker use of LEV and RPE, and cleaning and housekeeping practices.

Sampling methodology

Samples were collected using an air sampling pump (Sidekick pump; SKC Ltd, Dorset, UK.) with Higgins Dewell cyclone (Casella, Bedford, UK.) and 25mm, 5 µm pore size polyvinyl chloride filters pre-calibrated to a flow rate of 2.2 l min−1 with a primary air flow meter (DryCal DC Lite; BIOS International, NJ, USA). The sampling pump was attached to a belt around the waist or to a harness. The cyclones were attached to the worker’s lapel within 30cm of his breathing zone, ensuring the cyclone inlet was in a downward vertical position. Workers were asked to perform their work tasks as normal to ensure that the dust levels measured were representative of normal work activities. This included normal use of all tools, engineering controls, and RPE. Task sampling was performed, and when sampling a work task, all preparatory work i.e. marking out of stone, positioning of stone, the work task, and all clean-up activities after the task were monitored. Workers typically only performed one single task per work shift.

The respirable dust samples were analysed gravimetrically according to HSE MDHS 14/3, and the RCS content on the filter was quantified by X-ray diffraction as per HSE MDHS 101. All laboratory analytical analysis was carried out by the Institute of Occupational Medicine in Edinburgh. Samples below the analytical limit of detection (LOD) for crystalline silica were reported as <0.02mg.

In addition, real-time dust measurements were carried out using a SidePak AM510 personal aerosol monitor (TSI Incorporated, Shoreview, MN, USA) adjusted to measure respirable dust using a 10mm Dorr-Oliver cyclone attachment. The SidePak AM510 was calibrated to the recommended flow rate of 1.7 l min−1 using a primary air flow meter (DryCal DC Lite; BIOS International, NJ, USA.) and was set to log data at 1-min intervals. The data were downloaded to a computer using TSI Trackpro software. Results of the SidePak were used to examine exposure patterns whilst different tasks were carried out and to identify tools and tasks which created high levels of dust.

Statistical analysis

Task-specific exposure data were expressed as 8-h time-weighted average (TWA) concentrations. It was assumed that the RCS exposure for the remainder of the work shift was zero for the following reasons: the worker was carrying out non-stonework activities such as driving a forklift or transporting finished stone pieces to the monument, and the worker being monitored was working away from the stone workshop or any colleagues carrying out stonework. The exposure data were approximately log normally distributed and the geometric mean (GM) and geometric standard deviation of the 8-h TWA average exposure data were calculated.

For testing compliance with the Occupational Exposure limit value (OELV) the joint document by the British and Dutch occupational hygiene societies (BOHS and NVvA) on ‘Testing Compliance with Occupational Exposure Limits (OELs) for Airborne Substances’ (BOHS, 2011) was followed. Exposure data were categorized into similar exposure groups based on material and task and compliance was estimated by comparing the 95th percentile of the exposure distribution with the OELV (BOHS, 2011).

In order to investigate determinants of RCS exposure and their contribution to the individual’s mean exposure, mixed-effects modelling with forward inclusion of the variables was carried out using the log-transformed RCS 8-h TWA data for materials granite and sandstone (n = 65) as the dependant variable. RCS 8-h TWA data for materials limestone and lime mortar were not included in the model because of the large number of non-detect data. Material worked on, worker task, RPE, LEV, job title, weather, level of enclosure, and task duration were introduced as fixed variables, and three sources of random variance were looked at—between-depot variability, between-workers within-depot variability, and within-worker variability. RPE and LEV each had two levels depending on whether they were present or not. Enclosure had three possible levels (indoors, outdoors, and partial enclosure), while weather was grouped into six levels (wet and windy, dry and windy, wet and still, dry and still, showers and damp, and sunny). Forward stepwise regression was used to introduce variables significant at the P ≤ 0.001 level one at a time. Variables were added until no improvement to the model was made. The results of the regression modelling were reported as β, SE and Exp(β). β is the coefficient of the fitted line from the regression modelling, and SE is the standard error of the estimated β. As the regression modelling was carried out with data on the log scale, the back transformation of the β (Exp(β)) was also determined. In this context, the Exp(β) could be described as the ratio of the GMs and can be interpreted as the percentage increase (or decrease) in exposure associated with the factor (compared with the baseline factor level).

The extent of reduction in the between-depot, between-worker (within-depot), and within-worker variance components was also investigated. For results below the LOD, levels were imputed following a technique described by Lubin et al., (2004). This involved replacing below LOD values with randomly imputed values between zero and the LOD. A single imputation was carried out. All statistical analyses were performed using GenStat software (14th Edition; VSN International Ltd).
RESULTS

Exposure controls walk-through survey

All depots contained one stone cutting workshop which contained the following tools: a water-cooled primary cutting saw and hand tools including disc polisher/cylinder polisher; 5″, 9″, and 12″ angle grinders; pneumatic chisels; hand chisels; brushing tools; and hand punches, as well as a centralized LEV system. The LEV system included one or more movable extraction arms (Nederman Extraction Arm Original) (Nederman, 2010) connected at various locations around the workshop to centralized ducting. The ducting was connected to a Nederman L-PAK 250 compact stationary high-vacuum unit. The inlets used on the extraction arms were plastic with a hood diameter of 16cm. The filtered air was emitted to the external environment through a vent. The water-cooled primary cutting saw was a bridge saw with a 2 m blade. Water was applied to the blade via a recirculating tank at a rate of 100–120 l min−1.

Results of the walk-through survey indicated that there were control measures present in all depots, but they were regularly misused or inadequate for the task. Recommended guidance on engineering controls for stone masonry work involving power tools (HSE, 2006) were not complied with on any of the sites visited. A centralized LEV system was present in all depots, and although there was evidence of correct installation, this system was purchased without consultation with the workers or consideration of the work processes to be controlled. Issues regarding the use of the LEV system observed in all depots included the worker not working within the capture zone of the LEV and the capture arm being unable to deal with the high volumes of dust from tasks such as cutting and grinding.

The LEV systems in place, although working as per specification (capture velocity 11 m s−1, n = 12), were not suitable for the processes that were being carried out in the workshop. The tasks of grinding and cutting stone with a 5″, 9″, and 12″ grinder produced large clouds of dust-laden air too rapidly for it to be captured by the capture arm. Workers regularly worked with large pieces of stone up to 3 m in length, and it was not practical for the worker to regularly reposition the arm as this involved him stopping the task periodically. Capture arms are not recommended for these work tasks for this reason (HSE, 2001, 2011).

RPE was provided in all depots; however, the RPE used by workers varied widely and included nuisance dust masks, positive air purifying respirators, and disposable or reusable negative pressure RPE which in the majority of cases was an FFP3-disposable respirator or half-mask respirator with combination filters. Most RPE had an assigned protection factor of 20. However, there was no evidence of a workplace RPE program comprising of training, fit testing, and a formal purchasing policy for RPE.

Exposure measurements

A total of 103 exposure measurements were collected from 35 stoneworkers. Sampling times ranged from 30–375min with a median sampling time of 240 min. Repeated measurements were obtained for 19 workers, with the number of repeated measurements for each worker ranging from 2 to 16. Worker tasks measured included cutting stone on a water-cooled primary saw, cutting stone with 5″, 9″, and 12″ angle grinders, grinding stone with 5″ angle grinders, decorating stone with hand and pneumatic chisels, and repointing with a trowel. Stone materials worked on included sandstone, limestone, granite, and lime mortar. RPE was used by workers in 52% of the work tasks sampled. LEV capture hood systems installed were used for all grinding, cutting, and decoration tasks carried out in the stone workshops. Water suppression was used as the primary exposure control on the primary saw.

Figure 1 provides an example of results obtained with the SidePak photometer, which clearly shows the influence of three different tasks carried out by three different workers (grinding sandstone with a 5″ angle grinder in a partially enclosed workshop, cutting sandstone with a 5″ grinder in an enclosed work space, and repointing with lime mortar outdoors) on the dust exposure. During the grinding task, peaks in the dust measurements were due to the worker working outside of the capture zone of the LEV. During the cutting task, between the periods 11.13 and 11.53, the worker was not actively involved in stonework.

Table 2 presents a summary of the personal RCS exposure levels (mg m−3; 8-h TWA) grouped by material worked on and task. All active stonework and related activities carried out during the work shift were sampled and therefore 8-h TWA exposure levels presented are representative of full-shift exposures. RCS exposure for workers working with limestone and repointing with lime mortar (n = 38) ranged from <0.02 to 0.06mg m−3 (8-h TWA). RCS exposure for tasks involving granite ranged from 0.02 to 0.21mg m−3 (8-h TWA); 30% of these measurements exceeded the Irish OELV of 0.1mg m−3. RCS exposure levels for tasks involving sandstone ranged from <0.02 to 6.00mg m−3 (8-h TWA) with 57% of exposure measurements exceeding the Irish OELV of 0.1mg m−3. Highest RCS exposure values were recorded for the task of grinding (using a 5″ angle grinder; GM = 0.70mg m−3) and cutting (using 5″, 9″, and 12″ angle grinders) sandstone (GM = 0.70mg m−3), respectively (Table 2). Lowest RCS exposure values were recorded for the task of repointing with lime mortar (GM = 0.005mg m−3). Table 3 presents the percentage of measurements that exceeded the Irish OELV (0.1mg m−3; HSA, 2011), the occupational exposure standard recommended by the Scientific Committee on Occupational Exposure Limit Values (SCOEL) values (0.05mg m−3; SCOEL, 2002) and the threshold limit value (TLV) recommended by the American Conference of Governmental Industrial Hygienists (ACGIH; 0.025mg m−3; ACGIH, 2008).

For tasks involving sandstone, 67 and 76% of RCS exposure levels exceeded the SCOEL recommended OELV of 0.05mg m−3 and the ACGIH TLV of 0.025mg m−3, respectively. Fifty percent of tasks involving granite exceeded the ACGIH TLV of 0.025mg m−3, and 7 and 4% of measurements involving limestone and lime mortar, respectively, exceeded 0.025mg m−3 (Table 3 and Fig. 2).

Table 4 shows the coefficients of the fixed effects in the optimal model presented as β, SE, and Exp(β). Due to the natural nesting of workers within depots, three sources of random variance were looked at—between-depot variability, between-workers within-depot, and within-worker variability using mixed effects models (Table 5). After inclusion of the random terms of depot and worker in the model, the fixed effect that was found to be significantly associated with RCS exposure and therefore included in the model was task. Material (granite or sandstone), enclosure, RPE, weather, LEV, and task duration were not found to improve the model so were not included. Table 4 illustrates that the tasks of grinding and cutting result in average RCS exposures of between 32 and 70 times the exposures recorded for the task of decorating. No estimate was calculated for the task of repointing as this task was not carried out with sandstone or granite. The between-depot, between-worker, and within-worker variance components were reduced by 46, 89 and 49%, respectively, after including task in the mixed effects model (Table 5). The between-worker within-depot variance component was reduced from 0.80 to 0.09, after including task in the model, suggesting that the differences in exposure between workers (within a depot) was predominantly due to differences in tasks between the workers.
