3.1
Epidemiologic Considerations in Occupational Respiratory Disease Studies
3.1.1
Study Designs
Epidemiology
is the study of patterns of disease occurrence in human populations
and the factors that influence those patterns [Lilienfeld and Stolley
1994]. Epidemiology is the primary science used to study silica related
diseases in workers. Most epidemiologic studies of silica-exposed workers
discussed in this review are cross-sectional studies (i.e., prevalence
studies) or retrospective (i.e., historical) cohort studies. Cross-sectional
studies measure symptom or disease occurrence in a selected population
at one point in time. An example of a cross-sectional study design would
be the spirometric testing of lung function in a group of granite shed
workers during an annual health survey and comparison with respiratory
function in nongranite workers. Cross-sectional studies have two disadvantages:
- Usually
only the survivor population is examined. Retired, former,
or deceased workers are not included, possibly resulting in an underestimate
of the disease prevalence.
- It
may be impossible to determine whether exposure preceded the disease
if both are measured at the same time.
|
Construction
workers drilling holes in concrete pavement during highway repair.
Photograph by Kenneth Linch, NIOSH
|
Many epidemiologic
studies of silica-related diseases are retrospective cohort morbidity
or mortality studies. In this approach, the illnesses, deaths, and exposures
(surrogate or reconstructed) of an entire cohort (e.g., all workers
ever employed in one foundry) are followed forward from a time in the
past to a designated time in the future, and the number and causes of
deaths that occur in that interval are assessed. Exposures for the followup
period may be reconstructed from historical information or a surrogate
measure such as duration of employment. The mortality of the cohort
is then compared with the mortality of a standard population. For example,
Steenland and Brown [1995b] used a retrospective study design to examine
the mortality of a cohort of white male underground gold miners employed
for at least 1 year between 1940 and 1965. The miners were followed
from their first date of mining employment to their date of death or
until the end of 1990, whichever came first. Their mortality was then
compared with that of the U.S. population or the county where the mine
was located. A disadvantage of silicosis mortality studies that use
death certificate data is that silicosis cases could be under-ascertained
even when contributing causes of death are included, as suggested by
a study of silicosis mortality surveillance in the United States [Bang
et al. 1995].
3.1.2
Sources of Bias
Three
main (but not mutually exclusive) types of bias may affect the results
of epidemiologic studies of silica-exposed workersselection bias,
information bias, and confounding [Checkoway 1995]:
Selection
bias originates from the method of choosing study subjects.
This type of bias is a common criticism of lung cancer studies of compensated
silicotics because silicotic workers who sought compensation for their
disease may differ from all silicotics in symptoms, radiographic changes,
social and psychological factors, and industry [Weill and McDonald 1996;
McDonald 1995]. However, Goldsmith [1998] reviewed this question and
concluded that lung cancer risk estimates were not higher in compensated
silicotics when compared with those of silicotics ascertained from other
clinical sources (i.e., hospital or registry data).
Information
bias involves misclassification of study subjects by disease
or exposure status [Checkoway et al. 1989]. An example of disease (silicosis)
misclassification occurred in a study of North Carolina dusty trades
workers [Amandus et al. 1991; Rice et al. 1986]: a re-evaluation of
the chest X-rays found that 104 of the 370 cases categorized as silicosis
were actually International Labour Organization (ILO) category 0 (nonsilicotic)
[Amandus et al. 1992]. Sources of exposure assessment errors include
instrument error, incorrect imputation of exposure when data are missing,
and data extrapolation errors [Checkoway 1995]. Misclassification of
exposure may occur in retrospective cohort studies of silicosis when
quantitative dust exposure measurements are mathematically converted
from particle counts to gravimetric respirable silica equivalents.
Confounding
variables are factors that are related to exposure and
are also independent risk factors for the disease under study [Checkoway
1995]. Most studies of silica-related diseases controlled for confounding
factors such as age and race by study design or data analysis. Confounding
from cigarette smoking is an important concern in studies of lung cancer,
bronchitis, asthma, emphysema, chronic
obstructive pulmonary disease (COPD), and lung function. Confounding
of an exposure-disease relationship by cigarette smoking is less likely
when an internal comparison group is used e.g., when both groups are
from the same plant [Siemiatycki et al. 1988]. (Some studies in this
review used external comparison populations.) Most of the lung cancer
studies among underground miners did not control for the effects of
other carcinogens that may have been present, such as arsenic, radon
progeny, and diesel exhaust (see Section
3.4.1).
The effects
of bias discussed here can be minimized by applying epidemiologic methods.
Description of appropriate methodology is available in epidemiology
textbooks.
3.2
Silicosis
3.2.1
Definition
Silicosis
most commonly occurs as a diffuse nodular pulmonary fibrosis. This lung
disease (which is sometimes asymptomatic [NIOSH 1996b]) is caused by
the inhalation and deposition of respirable crystalline
silica particles (i.e., particles <10 µm in diameter) [Ziskind
et al. 1976; IARC 1987]. According to a report from the U.S. Surgeon
General [DHHS 1985], cigarette smoking has no significant causal role
in the etiology of silicosis. Probably the most important factor in
the development of silicosis is the dose of respirable silica-containing
dust in the workplace setting that is, the product of the concentration
of dust containing respirable silica in workplace air and the percentage
of respirable silica in the total dust. Other important factors are
- the
particle size,
- the
crystalline or noncrystalline nature of the silica,
- the
duration of the dust exposure, and
- the
varying time period from first exposure to diagnosis (from several
months to more than 30 years) [Banks 1996; Kreiss and Zhen 1996; Hnizdo
and Sluis-Cremer 1993; Hnizdo et al. 1993; Steenland and Brown 1995a;
ATS 1997].
Experimental
evidence supporting the influence of these factors has recently been
reviewed [Mossman and Churg 1998; Heppleston 1994]. Many in vitro studies
have been conducted to investigate the surface characteristics of crystalline
silica particles and their influence on fibrogenic activity [Bolsaitis
and Wallace 1996; Fubini 1997, 1998; Castranova et al. 1996; Donaldson
and Borm 1998; Erdogdu and Hasirci 1998]. These researchers found that
a number of features may be related to silica cytotoxicity. Further
research is needed to associate the surface characteristics with occupational
exposure situations and health effects [Donaldson and Borm 1998]. Such
exposure situations may include work processes that produce freshly
fractured silica surfaces [Bolsaitis and Wallace 1996; Vallyathan et
al. 1995] or that involve quartz contaminated
with trace elements such as iron [Castranova et al. 1997].
A worker
may develop one of three types of silicosis, depending on the airborne
concentration of respirable
crystalline silica:
- chronic
silicosis, which usually occurs after 10 or more years of exposure
at relatively low concentrations;
- accelerated
silicosis, which develops 5 to 10 years after the first exposure;
or
- acute
silicosis, which develops after exposure to high concentrations
of respirable crystalline silica and results in symptoms within a
period ranging from a few weeks to 5 years after the initial exposure
[NIOSH 1996b; Parker and Wagner 1998; Ziskind et al. 1976; Peters
1986].
The symptoms
of accelerated silicosis are similar to those of chronic silicosis,
but clinical and radiographic progression is rapid. Also, fibrosis may
be irregular and more diffuse [Banks 1996; Seaton 1995; Silicosis and
Silicate Disease Committee 1988] or not apparent on the chest radiograph
[Abraham and Weisenfeld 1997]. Acute silicosis is typically associated
with a history of high exposures from tasks that produce small particles
of airborne dust with a high silica content, such as sandblasting, rock
drilling, or quartz milling [Davis 1996]. The pathologic characteristics
of acute silicosis (sometimes referred to as silicoproteinosis) resemble
those of alveolar proteinosis [Wagner 1994; Davis 1996]. Pulmonary fibrosis
may not be present in acute silicosis [NIOSH 1996b].
Epidemiologic
studies of gold miners in South Africa, granite quarry workers in Hong
Kong, metal miners in Colorado, and coal miners in Scotland have shown
that chronic silicosis may develop or progress even after occupational
exposure to silica has been discontinued [Hessel et al. 1988; Hnizdo
and Sluis-Cremer 1993; Hnizdo and Murray 1998; Ng et al. 1987; Kreiss
and Zhen 1996; Miller et al. 1998]. Therefore, removing a worker from
exposure after diagnosis does not guarantee that silicosis or silica-related
disease will stop progressing or that an impaired workers condition
will stabilize [Parker and Wagner 1998; Weber and Banks 1994; Wagner
1994].
3.2.2
Epidemiologic Exposure-Response Models of Silicosis
This section
reviews published epidemiologic studies that provide evidence of an
exposure-response relationship for crystalline silica and silicosis
using cumulative exposure data. Exposure-response models based on cumulative
exposure data can predict silicosis risk for a particular silica dust
exposure over a period of time. Epidemiologic studies that provided
evidence of an exposure-response relationship for silica and silicosis
on the basis of other kinds of exposure data (e.g., duration of exposure)
have been reviewed elsewhere [EPA 1996; Davis 1996; Hughes 1995; Rice
and Stayner 1995; Seaton 1995; Steenland and Brown 1995a; Goldsmith
1994a; WHO 1986].
Table
12 summarizes the published studies that predict the incidence
or prevalence of radiographic silicosis based on models of cumulative
exposure to respirable crystalline silica. Table
13 presents details about the cohorts, quartz content of the dust,
followup periods, and limitations of each study. All of the studies
predicted the occurrence of at least one case of radiographic silicosis
per 100 workers at cumulative exposures approximately equal to the OSHA
and MSHA PELs and the NIOSH REL over a 40- or 45-year working lifetime
(see appendix for the PELs and REL).
Three studies predicted prevalences of 47% to 95% at the OSHA PEL. Each
study followed a cohort of miners for at least three decades from first
employment in the industry [Kreiss and Zhen 1996; Hnizdo and Sluis-Cremer
1993; Steenland and Brown 1995a]. Studies of foundry workers [Rosenman
et al. 1996], hardrock miners [Muir et al. 1989a,b; Muir 1991], and
workers in the diatomaceous earth industry [Hughes et al. 1998] followed
workers for less than 30 years (mean) and predicted prevalences of 1%
to 3%. The studies presented in Table 12
predicted that approximately 1 to 7 silicosis cases per 100 workers
would occur at respirable quartz concentrations
of 0.025 mg/m3half the NIOSH REL of 0.05 mg/m3with
the contingencies and exceptions noted in Table
12. However, that concentration
cannot be measured accurately at this time for the reasons given in
Section 2.4.
Table
12 does not include a cohort study of 1,416 coal miners exposed
to coal dust with quartz concentrations
ranging from 0.4% to 29.4% of respirable dust [Miller et al. 1998].
This study predicted pneumoconiosis risks for 47 men with a profusion
of median small opacities of ILO category
>2/1 (i.e., 2/1+), a higher category of radiographic
abnormality than reported in the studies listed in Tables
12 and 13. Logistic regression models
predicted that the risk of small opacities of 2/1+ at the time of followup
examination would be about 5% for miners exposed to a mean respirable
quartz concentration of 0.1 mg/m3
and about 2% for miners exposed to a mean respirable quartz concentration
of 0.05 mg/m3 for about 15 years [Miller et al. 1998]. The
predicted risks increased with cumulative exposure to respirable quartz
dust.
A currently
unpublished study of 600 retired Vermont granite workers found nodular
opacities consistent with silicosis (degrees of profusion not reported
in abstract) in 4.7% of 360 radiographs read by three readers [Graham
et al. 1998]. The average duration of employment for these workers was
31 years, and the average time from first exposure to radiographic examination
was 39 years. Most workers in the cohort were first employed after 1940,
when average quartz dust concentrations were below the current OSHA
PEL [Graham et al. 1991; Ashe and Bergstrom 1964].
Although
the variability in prevalence estimates (i.e., 1% to 90%) cannot be
solely attributed to differences in followup periods, chronic silicosis
is a progressive disease, and its development after a long latency period
and after workers leave employment must be accounted for in epidemiologic
studies. A study of autopsied gold miners in South Africa also supports
the need for examining workers after a long latency period and after
they leave employment [Hnizdo et al. 1993]. Radiologic findings for
profusion of rounded opacities (ILO category >1/1) were compared
with pathological findings for silicosis in 326 miners with an average
of 2.7 years between the radiologic and pathologic examinations. Silicosis
was not diagnosed radiographically for at least 61% of the miners with
slight to marked silicosis at autopsy. The probability of a false negative
reading increased with years of mining and average concentration
of respirable dust [Hnizdo et al. 1993]. Experimental studies of rats
also reported a lack of complete agreement between histopathologic indicators
of silica dust exposure and radiographic readings [Drew and Kutzman
1984a,b].
In addition,
improved exposure assessment methods and data analyses that account
for variations and deficiencies in exposure data would improve the risk
estimates for silica-exposed workers [Agius et al. 1992; Checkoway 1995].
Although epidemiologic studies that used cumulative exposure estimates
represent the best available source of information for characterizing
the dose-response relationship in occupational cohorts, peak exposures
may predict silicosis risk better than cumulative exposures [Checkoway
and Rice 1992]. However, data on peak exposures are rarely available,
and data supporting this hypothesis are limited.
3.3
TB and Other Infections
3.3.1
Definition
As silicosis
progresses, it may be complicated by severe mycobacterial or fungal
infections [NIOSH 1996b; Ziskind et al. 1976; Parkes 1982; Parker 1994].
The most common of these infections, TB, occurs when the macrophages
are overwhelmed by silica dust and are unable to kill the infectious
organism Mycobacterium tuberculosis [Parker 1994; Ng and Chan
1991; NIOSH 1992a,b; Allison and Hart 1968]. About half of the mycobacterial
infections that occur in workers with exposure to silica are caused
by M. tuberculosis, and the other half are caused by the nontuberculous
mycobacteria (NTM) Mycobacterium kansasii and Mycobacterium
avium-intracellulare [Owens et al. 1988; NIOSH 1996b]. Infections
in workers with silicosis may also be caused by Nocardia asteroides
and Cryptococcus [Ziskind et al. 1976; NIOSH 1996b; Parker 1994;
Parker and Wagner 1998]. ATS [1997] recommends that the diagnostic investigation
of a patient with silicosis and possible TB include consideration of
NTM disease. The ATS also recommends that tuberculin tests be administered
to persons with silicosis and to those without silicosis who have at
least 25 years of occupational exposure to crystalline
silica [ATS 1997].
3.3.2
Epidemiologic Studies
Recent
surveillance data indicate that TB rates in the United States are 5
to 10 times higher among racial and ethnic minorities (after adjustment
for the effects of age, sex, and country of birth) [Cantwell et al.
1998]. Cantwell et al. [1998] reported that the relative risk of TB
increased as socioeconomic status (measured by six indicators) decreased,
after adjustment for the effects of age (relative risks ranged from
2.6 to 5.6 in the lowest versus highest quartiles). The number of TB
cases among foreign-born persons in the United States increased by 56%
during the period 1986 to 1997 [CDC 1998c].
The association
between TB and silicosis has been firmly established by the results
of epidemiologic studies conducted during this century [Balmes 1990].
This association was supported by a survey of TB deaths among silicotics
in the United States for the period 1979 to 1991 [Althouse et al. 1995]
and by the results of four recent epidemiologic studies [Goldsmith et
al. 1995; Cowie 1994; Sherson and Lander 1990; Kleinschmidt and Churchyard
1997]. Black South African gold miners [Cowie 1994] and Danish foundry
workers [Sherson and Lander 1990] with chronic silicosis had threefold
and tenfold incidences of TB, respectively,
compared with nonsilicotic, non-silica-exposed workers of similar age
and race. Goldsmith et al. [1995] compared the mortality of 590 California
silicosis claimants with that of U.S. males and found that the TB mortality
of the claimants was 50 times that of all U.S. males (standardized
mortality ratio [SMR]=56.35; 45 deaths observed, 0.8 expected; 95%
confidence interval [CI]=41.1075.40).
A retrospective study of TB among 4,976 miners from the Freegold mines
in South Africa reported that the incidence
rate ratio for miners with silicosis (ILO category >1/1)
was 1.54 (95% CI=1.002.37) compared with miners without silicosis
(after adjusting for the effects of age, followup period, cumulative
service, and occupation) [Kleinschmidt and Churchyard 1997]. The incidence
of TB for the oldest age group was 21 times that of the youngest group
(incidence rate ratio=21.17; 95% CI=8.6052.11); and for workers
in occupations with high dust exposure (such as drilling), the incidence
was twice that of surface and maintenance workers (adjusted incidence
rate ratio=2.25; 95% CI=1.493.38) [Kleinschmidt and Churchyard
1997].
Some evidence
indicates that workers who do not have silicosis but who have had long
exposures to silica dust may be at increased risk of developing TB.
Two epidemiologic studies reported that, compared with the general population,
a threefold incidence of TB cases occurred among 5,424 nonsilicotic,
silica-exposed Danish foundry workers employed 25 or more years [Sherson
and Lander 1990], and nearly a tenfold incidence occurred among 335
nonsilicotic, black South African gold miners with a median underground
employment of 26 years [Cowie 1994].
Westerholm
etal. [1986] found 13 cases among 428 silicotic Swedish iron and steel
workers and 1 case in a comparison group of 476 Swedish iron and steel
workers with normal chest radiographs (level of statistical significance
not reported). Both groups had been exposed to silica for at least 5
years.
A study
of TB incidence in 2,255 white South African gold miners included 1,296
miners who had an autopsy [Hnizdo and Murray 1998, 1999]. The smoking-adjusted
relative risk for TB in miners without silicotic nodules on autopsy
examination (n=577) increased slightly with quartiles of cumulative
dust exposure (relative risk=1.38 [95% CI=0.335.62]
for the highest quartile of cumulative exposure). For miners without
radiologically diagnosed silicosis (n=1,934), the smoking-adjusted relative
risk increased to 4.01 (95% CI=2.047.88) in the highest quartile
of cumulative dust exposure [Hnizdo and Murray 1998, 1999]. The authors
defined radiologic silicosis as ILO category
>1/1. TB was diagnosed, on the average, 7.6 years after the
end of dust exposure and 6.8 years after the onset of radiological silicosisa
result that supports the need for medical surveillance of workers after
the end of exposure to silica dust [Hnizdo and Murray 1998]. Miners
who developed TB before completing 10 years of underground employment
were excluded because they were not allowed to continue working underground
after diagnosis.
Corbett
et al. [1999] conducted a recent case-control study of TB and pulmonary
disease caused by NTM in South African gold miners. These researchers
found that radiographic silicosis, focal radiological scarring, and
human immunodeficiency virus (HIV) infection were significant risk factors
for NTM disease and for TB. Past medical history of TB treatment (odds
ratio [OR]=15.1; 95% CI=7.6429.93) and current employment in a
dusty job at the mines (OR=2.5; 95% CI=1.464.44) were
significant risk factors for NTM. ORs for NTM and TB increased with
years of employment (range of ORs was 1.0 to 9.4 for NTM and 1.0 to
4.1 for TB). The study included 206 NTM patients and 381 TB patients
of known HIV status admitted to a South African hospital. Also included
were 180 controls who were HIV-tested surgical or trauma patients admitted
to the same hospital during the same period.
Two recent
studies about silica exposure and TB used U.S. occupational mortality
data to conduct a proportionate mortality study of persons with TB by
occupation for 1979 through 1990 [CDC 1995; Chen et al. 1997]. Although
the study design did not control for confounding, it identified six
occupational groups with potential exposure to silica dust that had
age adjusted proportionate mortality ratios
(PMRs) for TB that were statistically significant (lower bound of the
95% CI>100) or greater than 200. Table
14 shows significant PMRs by race for construction occupations,
mining machine operators, grinding and polishing machine operators,
furnace and kiln operators, laborers, and mixing and blending machine
operators [CDC 1995].
Chen et
al. [1997] conducted a case-control study (8,740 cases; 83,338 controls)
with U.S. National Occupational Mortality Surveillance (NOMS) data for
1983-1992. The study controlled for confounding from age, sex, race,
socioeconomic status, potential exposure to active TB, and the presence
of silicosis and other pneumoconioses. The potential for silica exposure
was based on data from NOES [NIOSH 1988] and the National Occupational
Health Survey of Mining (NOHSM) [NIOSH 1996c]. This potential was categorized
as high, intermediate, or low or no.
The study found that decedents with high potential for exposure to silica
and no documentation of silicosis on the death certificate had a 30%
greater odds of mortality from respiratory TB than decedents with no
potential exposure to silica after adjustment by logistic regression
for the possible confounders mentioned earlier (OR=1.3; 95% CI=1.141.48).
The results also suggest an exposure-response relationship between silica
exposure (in the absence of silicosis) and death from respiratory tuberculosis
[Chen et al. 1997].
|