Occupational Exposure to Crystalline Silica Dust in the United States, 1988–2003

ABSTRACT

The purposes of this study were a) to summarize measurements of airborne (respirable) crystalline silica dust exposure levels among U.S. workers, b) to provide an update of the 1990 Stewart and Rice report on airborne silica exposure levels in high-risk industries and occupations with data for the time period 1988–2003, c) to estimate the number of workers potentially exposed to silica in industries that the Occupational Safety and Health Administration (OSHA) inspected for high exposure levels, and d) to conduct time trend analyses on airborne silica dust exposure levels for time-weighted average (TWA) measurements. Compliance inspection data that were taken from the OSHA Integrated Management Information System (IMIS) for 1988–2003 (n = 7,209) were used to measure the airborne crystalline silica dust exposure levels among U.S. workers. A second-order autoregressive model was applied to assess the change in the mean silica exposure measurements over time. The overall geometric mean of silica exposure levels for 8-hr personal TWA samples collected during programmed inspections was 0.077 mg/m3, well above the applicable American Conference of Governmental Industrial Hygienists threshold limit value of 0.05 mg/m3. Surgical appliances supplies industry [Standard Industrial Classification (SIC) 3842] had the lowest geometric mean silica exposure level of 0.017 mg/m3, compared with the highest level, 0.166 mg/m3, for the metal valves and pipe fitting industry (SIC 3494), for an 8-hr TWA measurement. Although a downward trend in the airborne silica exposure levels was observed during 1988–2003, the results showed that 3.6% of the sampled workers were exposed above the OSHA-calculated permissible exposure limit.

MATERIALS AND METHODS

Data sources.

The OSHA Integrated Management Information System (IMIS) database was used for the analysis of the airborne concentration of crystalline silica exposure (OSHA IMIS 2003). The OSHA IMIS database contained personal sample measurements of silica exposure (n = 11,036) collected during 3,732 OSHA inspections conducted between 1988 and 2003. Of the 11,036 samples, 203 duplicate measures of personal samples were excluded because the number of personal silica samples exceeded the total number of workers who were sampled. A total of 3,188 samples with missing values and 436 area and bulk samples were excluded from the analysis. The remaining 7,209 personal samples collected during 2,512 OSHA inspections were used in this analysis.

Analytic methods.

The analytic framework used in this study is based on Stewart and Rice’s (1990) method for grouping industries with the highest geometric means and those with the lowest geometric means, where five or more samples were available. We selected a sample size of five arbitrarily as the minimum number required for obtaining stable and reliable descriptive statistics. Personal samples of silica exposure measurements were stratified into two groups by type of inspections to explore if estimates of silica samples were biased in any direction: a) all 2,512 inspections and b) 948 programmed inspections. Two separate estimation analyses were conducted. First, we analyzed all personal samples (n = 7,209) of silica exposure measurements collected during OSHA inspections to determine whether estimates of silica samples collected during complaint, referral, monitoring, follow-up, and fatality inspections were highly biased toward the upper end. Second, we analyzed only personal samples (n = 2,868) randomly collected during programmed inspections. In this later analysis, samples collected during complaint, referral, monitoring, follow-up, and fatality inspections were excluded.

In this article, the term "exposure" is defined as the concentration of airborne occupational crystalline silica dust measured in the workers’ personal breathing work environment. In this study we focused on the analysis of personal samples of silica exposure levels measured as an 8-hr TWA measurement among workers in various industries and occupations, and silica levels are expressed as milligrams per cubic meter.

The term "industry" is defined as a group of establishments that primarily engaged in the same kind of economic activity, regardless of their types of ownership. Industries were coded using four-digit Standard Industrial Classification (SIC; Office of Management and Budget 1987) codes. The term "occupation" is defined as a collection of jobs or types of work requiring similar skills, responsibilities, educational requirements, training, licensure and credentials, and the like, found within various industries. To update silica exposure levels among workers with different job titles, the high-risk industries of "stonework masonry" (SIC 1741) and "gray iron foundry" (SIC 3321) with exposure levels above ACGIH TLV of 0.05 mg/m3 were selected (Dear 1996). Using 1997 county business patterns (U.S. Census Bureau 1997) and reports to OSHA inspectors by the facility (OSHA IMIS 2003), the percentage and number of workers potentially exposed to crystalline silica by selected industries were estimated.

Airborne silica measurement.

Personal samples of airborne respirable silica particles were collected using OSHA method ID-142 for quartz in workplace atmosphere using a personal sampling pump and a cyclone assembly (OSHA 1996). Using this method, a respirable sample was collected by drawing air at approximately 1.7 L/min through a 10-mm nylon Dorr-Oliver cyclone attached to a 37-mm diameter polyvinyl chloride filter cassette with a 5-μm pore size (part no. 625413, Mine Safety Appliances, Pittsburgh, PA; or cat. no. P-503700, Omega Specialty Instrument Co., Chelmsford, MA). The cyclone assembly and sampling pump were placed on an employee to collect samples of tiny respirable silica particles from the air in the breathing zone of the employee during an 8-hr work shift. Samples were properly packaged and shipped to the OSHA Salt Lake Technical Center (OSHA 1996). The sample particulates were dissolved in tetrahydrofuran and analyzed by using X-ray diffraction. The qualitative limit of detection for quartz is 5 μg. Further laboratory details are available in OSHA (1996).

Statistical analyses.

Statistical analyses were conducted using SAS software (SAS Institute, Inc. 1999). First, we conducted a univariate analysis to examine the distribution of the airborne silica exposure levels. We used natural-log transformation of airborne silica exposure levels because silica levels had a positively skewed distribution. In addition to arithmetic mean and median, geometric mean of airborne silica levels and geometric standard deviation (GSD) were calculated for each industry over the period of 1988–2003. Second, the prevalence of elevated crystalline silica exposure levels for 8-hr TWA measurements among workers in high-risk industries and occupations was estimated. Third, a non-parametric regression was applied to make multiple comparisons of silica exposure levels among different major industries, and the null hypothesis of equal variances among different categories of industries and for significant differences in mean exposure levels among industries was tested using F-test statistics. Fourth, mixed autoregressive and moving average model [ARMA (1,1)] regression analyses were conducted to evaluate time trends in the silica exposure levels. Finally, a second-order autoregressive error model was created to regress exposure levels on time with errors from one period to be related to errors from the previous two periods. A finding of p ≤ 0.05 was considered statistically significant.

The covariates examined for association with higher airborne silica levels were industry, inspections, and year. Industries were grouped into four categories based on the four-digit SIC codes: construction (1521–1799), manufacturing (2011–3299, 3411–3999), metal (3312–3399), and service combined with all other industries including wholesale and retail trade and finance, insurance and real estate, and transportation, communication, and utility (4011–9721) (Office of Management and Budget 1987). Because mining and agricultural industries were not addressed in the OSHA IMIS data, both industries were excluded from this analysis. Dummy variables were used to adjust the significant effect of various industry groups.
RESULTS

Prevalence of elevated airborne crystalline silica in occupations and industries.

In the construction industry, "stonework masonry" (SIC 1741) that primarily engages in masonry work, stone cutting, bricklaying, and the like, has been one of the high-risk industries where overexposure to silica exists. Within occupations, the prevalence of elevated airborne silica exposure levels ≥ 0.10 mg/m3 among workers with the job title "masonry worker" in the stonework masonry industry was twice as high (6.9%) as the prevalence among workers with the job title "bricklayer" in the same industry (3.1%).

The prevalence of elevated airborne crystalline silica exposure levels ≥ 0.50 mg/m3 was 0.5% (n = 36) for all sampled workers (Figure 1). The proportion of workers with elevated airborne silica exposure levels ≥ 0.10 mg/m3 was 29.9% (n = 2,106). Within industries, workers in the metal industry had a prevalence of elevated airborne silica exposure levels ≥ 0.05 mg/m3 (35.6%), 2.9 times higher than the prevalence among workers in the construction industry (12.4%).

Airborne crystalline silica dust levels.

Table 1 presents arithmetic mean, geometric mean, standard deviations, and median of 8-hr TWA exposure measurements by industries with the highest and lowest airborne silica exposure. Geometric mean (GSD) airborne silica exposure levels were between 0.017 mg/m3 (GSD, 0.931 mg/m3; surgical appliances supplies industry, SIC 3842) and 0.166 mg/m3 (GSD, 0.943 mg/m3; metal valves and pipe fitting industry, SIC 3494). The geometric mean and GSD airborne silica exposure levels by industries and type of inspections are shown in Table 2. The overall geometric mean of silica exposure levels for samples collected during programmed inspections was 0.077 mg/m3. The geometric mean of samples collected under all inspections combined was higher in eight industries, whereas the geometric mean from programmed inspections was higher in two industries (Table 2).
Table 3 presents the airborne silica exposure levels by occupations in the "gray iron foundry" industry (SIC 3321). Gray iron foundry is the industry that primarily engages in manufacturing gray and ductile iron castings, including cast iron pressure and soil pipes and fittings. Workers with the job title "spruer" had the highest geometric mean airborne silica exposure levels (0.154 mg/m3), followed by workers with the job title "hunter operator" (0.093 mg/m3), those with the job title "charger" (0.091 mg/m3), and workers with the job title "core maker" (0.078 mg/m3).

The airborne silica exposure levels by occupations in the "stonework masonry" industry (SIC 1741) are shown in Table 4. The overall geometric mean of silica exposure levels for workers in this industry was 0.065 mg/m3. The geometric mean silica exposure levels were highest in those workers with the job title "helper" (0.099 mg/m3), followed by those with the job title "stone cutter" (0.070 mg/m3), those with the job title "bricklayer" and "laborer" (0.067 mg/m3), and workers with the job title "masonry worker" (0.065 mg/m3).

There were an estimated 119,381 workers potentially exposed to crystalline silica in 18 selected industries (Table 5). An estimated 25,027 workers were potentially exposed to airborne silica exposure in the automotive repair paint shop (SIC 7532) compared with 114 workers in the metal valves and pipe fitting industry (SIC 3494). Workers potentially exposed to silica exposure in stonework masonry (SIC 1741), plastering drywall work (SIC 1742), and tile, marble, and mosaic work (SIC 1743) were estimated at 44,989 employees. Workers in the testing laboratories services (SIC 8734) were estimated at 18,497 potentially exposed to airborne silica exposure.

The nonparametric regression showed the mean square error (MS) in airborne silica exposure between industries (MSbi = 0.048) and within industries (MSwi = 0.014), with F (3, 7,205) = 3.28 (p = 0.02). In this analysis we rejected the null hypothesis of no significant differences in the mean exposure levels between industries. We attempted to fit a mixed autoregressive and moving average model, ARMA (1,1), to the silica exposure data. A chi-squared value of 12.6 (p = 0.01) showed that we could not reject the hypothesis that the residuals are correlated. Thus, ARMA (1,1) was not an adequate model for silica exposure data. A final second-order autoregressive error model showed that a decline in the airborne silica exposure levels of 10.0% was observed per year between 1988 and 2003, but it was not statistically significant (p = 0.18, R2 = 0.0398). Within industries, the autroregressive error model AR(2) predicted that the construction industry has significantly lower airborne silica exposure levels (p = 0.10) during this time period. The findings also predicted that manufacturing industries have higher silica exposure levels than the metal industries, but it was not statistically significant at p ≤ 0.05. The estimated autocorrelation coefficients ρ1 and ρ2 were −0.153 and −0.082, respectively, with an estimated variance of error term of 0.014. The results showed that the negative effect of an OSHA inspection on the airborne silica exposure levels was estimated at β = −0.007, with p = 0.0319.
