INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Silicosis was one of the earliest recognized occupational diseases; it is well established that workers exposed to high levels of airborne crystalline silica can suffer severe and debilitating disease and even death as a result of such exposure. In the United States from 1968 to 1990, over 6,000 death certificates listed silicosis as the underlying cause of death (1). Although severe cases of silicosis have become increasingly rare as levels of occupational exposure have decreased, many workers continue to be exposed to crystalline silica. It is therefore important that the level of risk experienced by current workers be as well characterized as possible, and that this information be used in setting occupational standards. The current study was undertaken to assess radiographic evidence of silicosis among workers in the diatomaceous earth industry, who are exposed primarily to the cristobalite form of crystalline silica. The main goals of the study were to determine whether there is an exposure-response relationship with the risk of opacities on chest X-ray, and to estimate risk for workers exposed at levels similar to current and recent levels prevailing in the workplace.

Workers in this study were employed in one diatomaceous earth mining and processing facility in Lompoc, California; operations at the facility included extraction of the mineral from open-pit mines, crushing of the ore, and heating of the crushed ore at high temperatures (calcination). When extracted, the mineral exists primarily as amorphous (noncrystalline) silica; after heating, the product typically consists of 10 to 60% crystalline silica, primarily in the form of cristobalite.

The results presented here are from an epidemiologic study that included a mortality follow-up of the study cohort (2), which was an extension of a previous mortality investigation (3). The same estimates of quantitative exposure as in the second mortality follow-up were used. In the mortality studies, mortality for lung cancer and nonmalignant respiratory disease were increased for the overall study population, and there were trends in risk with estimated cumulative exposure (respirable dust and crystalline silica), although the excess risk for lung cancer was predominantly concentrated in the highest exposure categories for both indices.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cohort and Exposure Assessment

The study cohort, which has been described elsewhere (2), consisted of 2,342 Caucasian males employed in a single California plant for at least 1 yr, with some of that employment occurring between 1942 and 1987, and who had not had known asbestos exposure in previous employment elsewhere.

Estimates of exposure to respirable dust for individual workers were derived through the use of quantitative air-monitoring data, available since 1948, and company work histories; a full description of the methods used to reach these estimates has appeared previously (4). Estimates of concentration levels of respirable dust prior to 1948 were made by using relative rankings of exposure levels over time, based on documented changes in processes and dust-control measures (3). Individual exposures to crystalline silica (mg/m³) were estimated from job category, estimated respirable dust concentrations, and the estimated percentage of crystalline silica in the product. The percentages of respirable dust estimated to be crystalline silica for jobs involving exposures to natural, calcined, and flux-calcined diatomite were 3%, 20% and 60%, respectively. If the exposure was to a mix of these diatomite types, a weighted percentage corresponding to the estimated mix was used.

For each worker, estimates of exposure to respirable dust and crystalline silica were summed over the years worked, up to the time of the relevant chest radiograph used for analyses (see DATA ANALYSIS section), yielding cumulative exposures (mg/m³-yr) up to the time of chest radiography. Cumulative exposures were divided by total years employed at time of radiography to obtain average concentrations of exposure (mg/m³) for respirable dust and crystalline silica.

Mean respirable dust concentrations by year for the study cohort have been reported previously (4). Mean crystalline silica concentrations exhibited the same patterns over time as did respirable dust levels, with four distinct time periods. For 1932 to 1943, 1944 to 1953, 1954 to 1973, and 1974 to 1994, the mean crystalline silica concentrations for workers with a chest radiograph were approximately 0.90 mg/m³, 0.40 mg/m³, 0.15 mg/m³, and 0.10 mg/m³, respectively.

For each year, the percentage of each worker's exposure to respirable dust that consisted of crystalline silica was calculated (cumulative crystalline silica exposure for that year divided by cumulative respirable-dust exposure, multiplied by 100). The mean values over time exhibited a steadily increasing trend from approximately 25% in the early 1930s to approximately 35% in the early 1960s (indicating increased use of calcined products), after which they remained relatively constant.

Chrysotile asbestos was used at various times in two operations in the plant; moreover, maintenance workers could have been episodically exposed to asbestos (of unknown fiber type). Quantitative estimates of asbestos exposures (fibers/ml) were derived for all cohort members (5).

Smoking information, collected systematically beginning in the 1960s, was available for approximately half the cohort members. This information, when available, only permitted distinctions of ever and never cigarette smokers.

Chest X-rays

Starting in the 1930s, the company maintained a health-surveillance program that included a chest X-ray of most workers at time of hire and periodically throughout employment. The numbers of films available for individual workers varied widely, primarily depending on the length of time employed. Long-term workers generally had numerous films. However, participation in the chest X-ray program was not mandatory, and some long-term workers had only a few films available. Eighty-two workers who retired but remained in the area continued to have films taken after leaving employment, but there was no systematic program of following all retirees.

One of the authors (R.N.J.) examined all available chest X-rays for each worker (including postemployment films, when available), and selected the latest film of acceptable quality. These latest films were read independently by three experienced "B" readers according to the International Labour Organization (ILO) 1980 Classification System (6). A film was considered to be positive for opacities if at least two of the readers judged the film to be positive either for large opacities or for small opacities of profusion  1/0. For each worker with a positive latest film, the readers independently read all films, noting the dates of his last negative and first positive films. To summarize the dates of the earliest positive film for each worker, the median of the readers' dates was used. Not reading earlier films for workers whose latest film was judged to be negative reflected the assumption that all earlier films for such a worker, if read, would also have been judged to be negative.

Of the 2,342 workers, 1,983 (84.7%) had X-ray films available. Availability of films varied by decade of hire, and was 77%, 79%, 86%, and 91% for workers hired < 1940, during the 1940s, during the 1950s, and 1960 or later, respectively. The quality of the films was generally good; only 0.3% (five of 1,983) were judged by median reading to be unreadable, and 13.5% (268 of 1,983) were judged to be of poor quality but readable.

Of the 1,978 workers with a readable film, 169 had only one film made at the time of hire (at or before hire or up to 1 mo after hire). These films were included in the readings and provided information about background rates of X-ray changes, but not information about the development of opacities after hire. Therefore, exposure-response analyses included only the 1,809 men with a film taken more than 1 mo after hire.

Data Analysis

Exposure indices were calculated from time of hire until the date of the X-ray film used for analysis. This was the date of the latest available film for workers for whom this film was negative for opacities, or the date of the first positive film for those for whom the latest film was positive for opacities.

Poisson regression analysis, done with generalized linear interactive modeling (GLIM) (7), was used to assess factors potentially related to the rates at which workers developed opacities on chest X-ray films. For each worker, the person-years at risk were calculated (from his time of hire until his time of the X-ray film used in the analyses) and allocated to specified categories of the factors being considered (e.g., age, cumulative crystalline silica exposure). As a worker aged and accumulated exposure, his person-years were allocated to changing categories for these variables, using the person-years program (8). For each category, the incidence of opacities was calculated as the number of workers in the category who developed opacities, divided by the person-years at risk.

Life-table analyses and parametric failure-time models were also implemented, using the SAS (9) LIFETEST and LIFEREG procedures (SAS Institute, Cary, NC), to assess cumulative risk of opacities on chest X-ray in relation to cumulative exposure. The log-rank test was used to test for statistically significant differences between survival curves. In fitting the failure-time models, exponential, Weibull, gamma, lognormal, normal, and loglogistic models were compared (10).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In reading the latest chest radiographs, made after the date of hiring, the three readers judged relatively similar percentages of workers (5.4%, 6.2%, and 4.1%) to be positive for opacities (Table 1). Using median (majority) reading, the prevalence was 4.5% (81 of 1,809). The prevalence of pleural changes (by median reading) was 1.8% (33 of 1,809). Of the 169 workers with only a film made at hire, three (1.8%) were positive for opacities by median reading; their ages at hire ranged from 39 to 56 yr.

Of the 81 workers with a film positive for opacities by median reading, 77 had small opacities of profusion  1/0 (with or without large opacities), and four had large opacities only. Of the 77 with small opacities, 26 had profusion 1/0 and 51 had profusion  1/1. Of the 81 workers with a positive film, 76 had an earlier film judged to be negative. For the remaining five workers, the earliest readable film was judged to be positive. These five workers were similar to the other 76 workers in terms of mean time between dates of hire and earliest film (11.3 and 11.4 yr, respectively), and were somewhat younger at hire (25.8 versus 30.9 yr); it was assumed that these five workers were negative for radiographic changes at the time of hire.

Of the 77 radiographs positive for small opacities of profusion  1/0, 25 (32%) were classified (by two or three readers) as having irregular small opacities as the primary type (of these 25 radiographs, 21 had irregular opacities as both the primary and secondary type). Considering only the 51 films positive at profusion 1/1 or greater, for which confounding by smoking-induced and age-associated irregular small opacities should have been minimal, 14 (27%) still showed irregular opacities as the predominant or exclusive type. Both types of small opacities were considered to be consistent with an effect of exposure to diatomite dust (pneumoconiosis).

Overall, the elapsed time from hire to making of the study radiograph (the latest film for those judged negative; the earliest positive film for those judged positive) ranged from 1 mo to 46 yr, with an overall mean of 11.5 yr; mean values were similar by decade of hire. The estimated mean concentration levels for respirable dust and crystalline silica were 0.93 and 0.30 mg/m³, respectively. These mean values differed substantially by decade of hire, ranging from 0.76 mg/m³ for exposure to crystalline silica for workers hired before 1940 to 0.12 mg/ m³ for those hired in 1960 or later. Although mean years employed at time of making of the study film were similar by decade of hire, there were substantial differences in the percentages of workers developing opacities on chest X-ray films, which were 23% (35 of 151), 7% (35 of 482) and 1% (11 of 1,176) for those hired before 1940, between 1940 and 1949, and in 1950 or later, respectively. For pleural changes, the corresponding percentages were 4.6%, 2.9%, and 1.0%, respectively. Of the 11 workers hired in 1950 or later who developed opacities, eight had profusion of 1/0 for small opacities and none had large opacities.

Crystalline Silica Exposure

There were increasing trends in the incidence of opacities with the categories both of age and cumulative exposure to crystalline silica. For the age categories 20 to 34, 35 to 44, and  45 yr, the incidence rates were 244, 506, and 585 (cases of opacities on chest X-ray per 100,000 person-years at risk), respectively. For cumulative crystalline silica exposures of  1, 1 to 3, 3 to 4, 4 to 6, 6 to 8, and > 8 mg/m³-yr, the incidence rates were 63, 259, 792, 1,319, 1,856 and 2,204, respectively. However, the trends in incidence rates across cumulative-exposure categories varied by age group, with the steepest trend observed in the youngest age group. Thus, for example, for the highest category of cumulative exposure to crystalline silica (> 8 mg/m³-yr), the incidence rate for the < 35 yr of age category was 4,027 cases per 100,000 person-years, compared with 1,020 for the 45+-yr age group. Poisson regression analysis confirmed a statistically significant interactive effect of age and cumulative exposure on risk of opacities (p < 0.0001). Because workers could enter high-cumulative-exposure categories at relatively young ages only by exposure to relatively high concentration levels of crystalline silica, these results suggest a role of concentration of exposure after accounting for cumulative exposure in the development of opacities on chest radiograph.

To investigate the role of exposure concentration of crystalline silica on the incidence of opacities, workers were classified into two categories based on the average concentrations of crystalline silica to which they were exposed (from hire to final or first positive chest X-ray): those with exposure to concentrations 0.50 mg/m³ and those with exposure to concentrations > 0.50 mg/m³. The value of 0.50 mg/m³ was chosen because the 357 workers who had a cumulative exposure to crystalline silica exceeding 3 mg/m³-yr (and who had a relatively high incidence of opacities, especially for the youngest workers) had been exposed to a mean crystalline silica concentration of approximately 0.50 mg/m³. Further subdivisions for workers with exposure to concentrations  0.50 mg/m³ were examined, but no differences were observed across these groups with respect to risk of opacities in relation to cumulative crystalline silica exposure; therefore, only the results for the  0.50 mg/m³ and > 0.50 mg/m³ groupings are presented. For these concentration categories (separately and combined), incidence rates and adjusted relative risks (RRs) by cumulative crystalline silica-exposure categories are summarized in Table 2. The percentages of workers developing opacities differed substantially in the two concentration groups, at 1.6% (23 of 1,452) and 16.2% (58 of 357), respectively. Moreover, for each cumulative crystalline silica-exposure category, the incidence rate and age-adjusted RR of opacities were higher for workers in the higher concentration category (> 0.50 mg/m³) than for those in the lower category. Poisson regression showed that for each of the two concentration groups separately, age was a statistically significant factor (p < 0.01), and after age had been accounted for, cumulative exposure was significant (p < 0.05 and p < 0.001 for the lower and higher concentration categories, respectively). After accounting for both age and cumulative crystalline-silica exposure, there was a significant difference in risk between the two concentration groups (p < 0.0001).

Survival Analysis

Survival analysis was used to further compare the risk of opacities on chest radiographs for the two groups of workers classified according to the concentration of crystalline silica to which they had been exposed, and also to calculate the cumulative risk of opacities by cumulative crystalline silica exposure. As with the results of the Poisson regression, there was a statistically significant difference in the survival distributions of these two groups (log-rank test, p = 0.0001). The loglogistic model provided the best fit to the data (followed closely by the Weibull model). Using this model, the estimated cumulative risk of having a positive radiograph was:

(1)

where x is cumulative crystalline silica exposure in mg/m³-yr and con is an indicator variable taking on the value 0 for a crystalline silica-exposure concentration  0.50 mg/m³ and 1 for an exposure concentration > 0.50 mg/m³. In fitting the model, the coefficient of the concentration indicator variable (con) was statistically significant (p < 0.0001).

The observed and fitted cumulative risks for the two exposure-concentration categories appear in Figure 1. For a cumulative crystalline silica exposure of 2.0 mg/m³-yr, the observed cumulative risks of X-ray opacities were 1.1% and 3.7% for the lower and higher concentration groups, respectively; for a cumulative exposure of 4.0 mg/m³-yr, the corresponding risks were 3.3% and 12.4%.

View larger version (16K): [in this window] [in a new window]   Figure 1.   Kaplan-Meier (KM) and fitted loglogistic cumulative risk of radiographic opacities in relation to cumulative crystalline silica exposure for two groups of workers based on mean concentration level of crystalline silica exposure. Solid and dotted lines represent the KM and fitted risks, respectively, for the high concentration group (> 0.50 mg/m3); dash/dot and dashed lines, respectively, represent the low concentration group ( 0.50 mg/m3). The numbers of workers entering each cumulative-exposure category (e.g., 0.0, 1.0, 2.0, 4.0, 6.0 mg/m3-yr) are indicated.

Early versus Late Hires

Not unexpectedly, estimated average concentration of exposure and period of hire were highly correlated. For example, of the 357 workers with an average crystalline silica-exposure concentration > 0.50 mg/m³, 319 (89%) were hired prior to 1950; of the 1,452 workers with an average exposure concentration  0.50 mg/m³, 1,138 (78%) were hired in 1950 or later. Consequently, analyses comparing results for workers by period of hire (< 1950 and 1950) found results closely similar to those comparing results by concentration category, including a statistically significant difference in the survival distributions of these two groups (p = 0.002 by the log-rank test). For the 1,176 workers hired  1950, the observed cumulative risks of X-ray opacities were 1.0% and 2.3% for cumulative crystalline silica exposures of 2.0 and 4.0 mg/m³-yr, respectively.

Risk was also examined in relation to duration of employment, without regard to cumulative crystalline silica exposure. For 40 yr employment, the cumulative risks were 0.048 among the 1,176 workers hired since 1950 and 0.307 for the 633 hired earlier.

Respirable Dust Exposure

Cumulative respirable dust exposure and cumulative crystalline silica exposure were highly correlated (Pearson's r = 0.93 for the logarithms of these variables). For all workers combined (without regard to mean crystalline silica-concentration level or period of hire), there was a statistically significant (p < 0.0001) trend in the age-adjusted RRs across categories of both exposure indices (Table 3). Although the highest RR was observed for the highest respirable dust category, risk exhibited a more steadily increasing trend with crystalline silica categories than that with respirable dust.

To determine whether there was an effect of crystalline silica exposure on opacity risk after accounting for dust, workers were dichotomized by percentage silica exposure, using the approximate mean percentage (i.e., < 35% versus  35%). In a life-table analysis of risk of opacities, after accounting for respirable dust, there was a statistically significant difference in the two percentage silica groups (p < 0.02 by the log-rank test); for each category of cumulative respirable dust, workers in the higher percentage silica group experienced a higher risk than those in the lower percentage silica group.

Asbestos Exposure

Although the greatest documented use of asbestos in the plant occurred in the 1950s and 1960s, the percentages of workers hired in 1950 or later who developed opacities or pleural changes on chest radiographs were lower than for earlier hires. Overall, of workers estimated to have had no asbestos exposure in the plant, 2.6% developed pleural changes and 7.5% developed opacities; of those with asbestos exposure, the corresponding percentages were 1.3% and 2.3%, respectively. More detailed analyses of radiographic changes (opacities or pleural changes) in relation to the quantitative asbestos-exposure estimates for individual workers found no relationships (results not shown).

Cigarette Smoking

Smoking information (ever smoker or not) was available for 71% of workers with an average crystalline silica-exposure concentration  0.50 mg/m³. Among these workers, smoking was significantly related to opacities, with 0.4% (1 of 269) of nonsmokers developing opacities, compared with 2.6% (20 of 756) of smokers (one-tailed p = 0.01). Smoking information was not generally available for the early workers, who tended to be those exposed at higher concentration levels. However, in previous reports (2, 3) only a minimal correlation between smoking and exposure was observed, after adjusting for calendar year; an unrealistically high correlation would be needed for smoking to account for the observed relations between opacities and cumulative exposure.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The risk of silicosis in diatomite workers exposed to ores from the same deposits as those in this study was initially reported in 1932 (11). The various degrees of severity of silicosis (almost 70% of 118 workers had some degree of disease) were related to duration of employment, and a number of recommendations to control dust and continue medical surveillance of the workers were made by the authors of the 1932 study. A later study by the U.S. Public Health Service (USPHS) (12) in five diatomite-production plants in three western states found that slightly less than 10% of the 869 workers in the study had radiographic evidence of silicosis, and that a similar percentage had radiographic changes considered more difficult to interpret. Of workers employed for more than 5 yr, 25% had radiographic evidence of silicosis, and nearly 50% of those employed at sites where the cristobalite content of the dust was high had such evidence. Workers exposed to natural diatomite (primarily amorphous silica) in the quarries had a much lower percentage of positive films than those exposed to the dust of calcined products containing a high percentage of cristobalite. Assessment of dust exposure in these plants found that the most recent measurements had shown substantial reduction of dust levels over previous levels; ongoing dust control, personal respiratory protection, and medical monitoring were urged.

Subsequent follow-up was done of employees of the California diatomite plant, with the follow-up population comprising 68% of the USPHS study population (13, 14); this plant was the same plant as in the present study. In the first of these follow-up studies (13), the prevalence of radiographic findings of pneumoconiosis had decreased to about 3%, with the decrease attributed to better dust control and the mandatory use of respirators by workers significantly exposed to cristobalite-containing dust. In the more recent report (14), definite evidence of silicosis (defined as profusion of small opacities of  1/1) was not detected in any worker employed for less than 25 yr (i.e., hired after 1955).

A recent cross-sectional study (15) of workers who were actively employed in the late 1980s in the same California plant as in the present study reported somewhat higher prevalences of small opacities than observed in the present study (26 of 446, or 5.8%, with profusion of 1/0 or higher). No information on participation rates was provided. A higher percentage of workers employed for more than 12.5 yr had small opacities than did those employed for less time (10.7% versus 1.0%).

Previous reports (14, 16) have indicated that nodulation on the chest radiograph is less discrete in pneumoconiosis from diatomite than from quartz, and the findings in the present study support this. Moreover, in the present study, the pneumoconiotic large opacities were notable for being "ill-defined" or poorly outlined, although this descriptor is no longer part of the ILO classification. This set of radiographs is thus recognizably different from series showing classical (i.e., purely nodular) silicosis that is the result of exposure to quartz.

In the present study, the high proportion of films with small opacities that also showed pneumoconiotic large opacities (28 of 77, or 36.4%) is noteworthy, since large opacities reflect fibrotic masses. Fibrosis does not appear to develop, in either humans or experimental animals in the absence of crystalline silica exposure. The epidemiologic and experimental evidence, including the results of this study, indicates that crystalline silica, whether quartz or cristobalite, is more likely than amorphous silica to produce radiographic evidence of fibrosis. The appearance of the scarring caused by crystalline silica is apparently altered by concurrent exposure to amorphous natural diatomite. This is analogous to so-called "mixed-dust fibrosis," in which exposure to quartz and iron oxide in hematite miners, for example, alters the histologic and radiographic appearances of scarring caused by quartz (17, 18). However, unless the "inert" dust alters the quantitative relationship between quartz exposure and fibrotic response, the distinction is merely descriptive in the visual sense, and the adverse health effect is, in fact, silicosis. The same reasoning should apply to diatomite, absent evidence for alteration of the crystalline silica-fibrosis relationship. Consequently, the choice of calling this disease silicosis or diatomite pneumoconiosis is only a matter of labeling.

In the present study, risk of opacities on chest radiographs was related to cumulative crystalline silica exposure, cumulative respirable dust exposure, concentration of exposure, decade of hire, age, and cigarette smoking. Trends in RR increased more consistently with cumulative crystalline silica exposure than with cumulative dust exposure, although the highest RR was observed for the highest respirable dust category. After accounting for respirable dust category, percentage-crystalline silica category (below or above 35%) was a significant factor in risk of opacities, supporting the conclusion that the opacities were a result of exposure to crystalline silica rather than to dust. In accord with the previous reports discussed earlier, the development of opacities was overwhelmingly concentrated in workers hired before 1950: 11.1% (70 of 633) of these earlier workers developed opacities, compared with 0.9% (11 of 1,176) of later workers.

The concentration of crystalline silica to which workers were exposed was an important risk factor for radiographic opacities, in that risk of opacities increased more steeply with cumulative crystalline silica exposure for workers exposed to high concentrations of crystalline silica (> 0.50 mg/m³) than for those exposed to lower levels. Although these results suggest that the concentration of crystalline silica to which exposure occurred was itself a risk factor, a possible alternative explanation is an underestimation of exposure levels for early hires (who constituted the great majority of the high-concentration group). The observed differences in risk between the workers exposed to high and low concentrations (and between early and late hires) for the same estimated cumulative crystalline silica exposure could have occurred if there was an underestimation of the exposure levels for early workers. For example, the observed cumulative risk was approximately 10% for a cumulative crystalline silica exposure of 8.0 mg/m³-yr for workers hired after 1950 and of 4.0 mg/m³-yr for workers hired before that date. Thus, if early levels of exposure to crystalline silica were in fact double the estimated levels, then this might account for much of the difference observed in exposure- response trends between the early and late hires.

As with most occupational epidemiologic studies, especially those in which the outcome of interest may not be detectable until many years after first exposure, one of the potential limitations of the current study was that workers largely participated in the routine chest X-ray program only during their employment. Therefore, development of opacities on radiographs subsequent to their leaving employment would not have been recorded for these workers. Although on average the latest available chest radiograph was made slightly more than 11 yr after the start of employment, there were 394 workers who had their latest films taken more than 20 yr after the date of their hire (ranging up to 50 yr, with a mean of 30.3 yr). Thus, the limitation is primarily one relating to the development of opacities more than 35 yr after hire, since only limited information was available for this period. It should be noted that loss to follow-up of workers after cessation of their employment may, but does not necessarily, lead to underestimation of cumulative risk. Estimates of the cumulative risk of developing opacities were based on the observed risks for workers entering increasing categories of exposure. Cumulative risk would have been underestimated only if there had been a healthy-worker effect, in which workers who left employment subsequently experienced a greater risk of developing opacities than those who continued employment.

An important issue is the level of potential risk to current workers in the diatomaceous-earth industry, who are exposed at levels of crystalline silica far lower than those of the early workers in this cohort. We believe that estimates of such risk based on this investigation should be based on the experience of workers hired since 1950, or, alternatively, those exposed to lower concentration levels ( 0.50 mg/m³). Exposure levels for both of these groups, are more representative of contemporary crystalline silica exposure levels; moreover, for workers hired after 1950, exposure data were available throughout their employment careers, without the need for extrapolation into the past.

Although knowledge of the existence of silicosis dates back many years, data addressing quantitative dose-response relationships for silicosis are limited and quite variable across studies. This variability could be due to differences in a number of factors, including: ascertainment of disease (e.g., death certificates; radiographic evidence for which differing definitions were used for the presence of opacities); availability of exposure data; and differing methods used to measure airborne crystalline silica dust.

Although the present study of diatomite workers examined potential risks of exposure primarily to the cristobalite type of crystalline silica, a number of studies have been done of workers exposed to the quartz polymorph. Based primarily on results from animal studies, current U.S. Government regulations deal differently with the permissible exposure limits (PEL) for these two polymorphs of crystalline silica; the PELs (for time-weighted averages) of the Occupational Safety and Health Administration (OSHA) are 0.1 mg/m³ for quartz and 0.05 mg/m³ for cristobalite.

Although a thorough review of the epidemiologic literature regarding exposure-response relationships for exposure to crystalline silica cannot be done here, it is of interest to compare the results of the present study with those of two others that made use of chest radiographs and individual work histories (19). In the present study, the cumulative risk of opacities was approximately 1.0% for a cumulative crystalline silica exposure of 2.0 mg/m³-yr (based on workers hired in 1950 or later). The cumulative risk (of small opacities of profusion  1/1) observed for Ontario hardrock miners (19, 20) for the same amount of cumulative exposure was 0.4% (based on a median of five readers). These results appear to be reasonably similar, especially since the definition of opacities in the present study included large opacities and small opacities of profusion  1/0. However, the cumulative risk of opacities (profusion  1/1) was reported to be considerably higher for South African gold miners, at approximately 10% for a cumulative dust exposure of 7.0 mg/m³-yr, corresponding to approximately 2.1 mg/m³-yr of crystalline silica exposure (21).

Other studies that have attempted to assess risk of silicosis in relation to quartz exposure (22) have used various study designs, and most had a number of limitations. These include a lack of information about past concentration levels; possible selection bias in a community-based investigation (27); exposures to other dusts, including asbestos (26); and primary ascertainment of silicosis by death certificates (24), which is recognized to have substantial uncertainties.

A cumulative risk of opacities of approximately 1.0% for a cumulative crystalline silica exposure of 2.0 mg/m³-yr, as observed in this study for later workers, is not appreciably higher than that reported for many unexposed populations. Two reviews of the literature on irregular small opacities (28, 29), although mainly concerned with interactions of age, smoking, and occupational exposures, implied substantial prevalences of idiopathic small opacities in nonexposed groups. A recent review (30) incorporated a metaanalysis of prevalence rates of small opacities (profusion  1/0) reported for nonexposed populations. For men in the North American studies included in this review, an overall pooled prevalence of 1.3% (95% confidence interval [CI] of 0.3% to 2.4%) was reported.

Most studies of workers exposed to crystalline silica-containing dust involve quartz exposure. The present study of risk of radiographic opacities consistent with silicosis in diatomite workers provides the only quantitative data on silica dose- response effects for cristobalite exposure. Our results do not support the view that for comparable crystalline silica exposures, cristobalite is more fibrogenic than quartz, as has been assumed in establishing U.S. Government exposure regulations. Nonetheless, our findings are consistent with an exposure- response relationship for silica, and therefore reinforce the need to enforce occupational exposure standards for protecting the health of workers, as well as the use of respiratory-protection measures when concentrations of silica may exceed allowable levels.