Cassidy, Adrian*; Mannetje, Andrea't†‡‡‡; van Tongeren, Martie‡; Field, John K.*; Zaridze, David§; Szeszenia-Dabrowska, Neonila¶; Rudnai, Peter∥; Lissowska, Jolanta**; Fabianova, Eleonora††; Mates, Dana‡‡; Bencko, Vladimir§§; Foretova, Lenka¶¶; Janout, Vladimir∥∥; Fevotte, Joelle***; Fletcher, Tony†††; Brennan, Paul‡‡‡; Boffetta, Paolo‡‡‡
In 1997, the International Agency for Research on Cancer (IARC) classified occupational exposure to crystalline silica as carcinogenic to humans.1 This decision was somewhat controversial, not least because studies of silica-exposed workers lacked exposure-response data and not all studies were consistent (with some indicating no association between exposure to silica and an increased risk of developing lung cancer2,3). In addition, previous studies have received criticism for not adequately taking into account potential confounding factors, including smoking and, to a lesser extent, exposure to other carcinogens, such as asbestos and polycyclic aromatic hydrocarbons.4 Using a weight-of-evidence approach similar to that used by the IARC, Hessel et al5 argued against the view that crystalline silica was carcinogenic in humans. Steenland3 has suggested that the lung carcinogenicity of silica has been controversial and difficult to establish because of its low relative potency. It is generally accepted that silica is a relatively weak carcinogen2; otherwise, the evidence for lung cancer would be far clearer and more convincing.6
Silica exposure occurs in a wide range of industries and occupations, including the “dusty trades” (such as mining and quarrying), potteries or ceramics, foundries, and various tasks in construction and manufacturing.7 Epidemiologic studies on the relationship between silica exposure and lung cancer have been primarily industry-based and include various ore mining and quarries, granite production, potteries, and diatomaceous earth industries.8–11 Industry-based studies generally have the advantage that exposure information is more readily available. However, many lacked detailed information on tobacco smoking and other potential confounders. There have been relatively few large community-based studies of silica and lung cancer, even though a significant proportion of workers in the general population are exposed to silica, albeit in most instances at lower concentrations than individuals employed in the dusty trades. One rarely studied exception is the construction industry, despite relatively high silica exposure levels (particularly for individuals engaging in masonry work, stonecutting, bricklaying or equivalent trades).12,13 In addition, few epidemiological studies have attempted to quantify respirable crystalline silica resulting from exposure to primary construction materials including sand, concrete, cement and brick dust.
The INCO Copernicus study is a large multicenter case–control study of occupational risk factors conducted in 7 European countries to investigate the role of occupational exposures in the development of lung cancer. Here, we report results of the association between occupational exposure to crystalline silica and risk of lung cancer, taking into account potential confounding factors, including smoking and concomitant exposure to other occupational carcinogens.
This case–control study was conducted in 15 areas in 6 Central and Eastern Europe countries (1 area in Romania and Russia, 2 in Poland, 3 in Slovakia and Czech Republic, and 5 in Hungary) and in one area in United Kingdom. Newly diagnosed cases of lung cancer were recruited from 1998 through 2002, as were the controls who were frequency matched to cases on sex and age (±3 years).
All persons ages 20 to 74 years with incident cases of histologically or cytologically confirmed lung cancer were included. Lung cancer included cancer in any of the topographic subcategories of code C34 of the International Classification for Diseases for Oncology (9th Revision). Identification of cases occurred through an active search of clinical and pathologic departments. Eligible cases were selected among hospitalized individuals, within 3 months of diagnosis: as a result, only a minority of cases died before interview and next-of-kin interviews were not required. An environmental exposure component of this study specified that cases lived in the study area for at least 1 year before diagnosis.
Most centers recruited hospital controls, whereas in 2 centers (1 in Poland and 1 in the United Kingdom) population controls were selected. Hospital controls were recruited either in the same hospitals or in general public hospitals serving the same areas from which the cases arose. Given that this study was designed to test multiple lung cancer hypotheses, hospital controls were selected from a prespecified list of diseases that excluded other cancers and diseases related to tobacco. No single disease represented more than 10% of the diagnoses of controls. Population controls were selected from population registers in Poland and from registers of General Practitioners in the United Kingdom.
A face-to-face interview was conducted by trained interviewers using a questionnaire assessing lifestyle factors and occupations held for more than 1 year. The lifestyle questionnaire collected detailed information on socioeconomic and demographic characteristics, medical history, family history of cancer, tobacco smoking, and lifetime residential and occupational history. The first stage of the occupational interview involved a general questionnaire, which elicited a complete occupational history and additional information relevant to the exposure assessment, including tasks performed, machines used, the work environment, time spent on each task and specific exposures. For 18 prespecified jobs, a specialized questionnaire also was completed for employment in any of the following jobs or industries: motor vehicle mechanic, wood worker, painter, welder, toolmaker or machine tool operator, miner or quarryman, insulation worker, meat workers, farmer, and the chemical, coke manufacture, foundry, glass, printing, rubber, steel, tanning, and asbestos compounds industries. These activities were selected based on likely exposure to 1 or more of the 70 agents, expected high prevalence in the study areas and inadequacy of the general questionnaire to capture the relevant information.
Local experts in industrial hygiene were selected in all study areas, with practical experience of industries in that region. The protocol was based on that used by Siemiatycki et al14 and adapted for the INCO Copernicus Study. Industrial activity of the employer and occupation of the study participant were coded by the experts using the “Statistical Classification of Economic Activities in the European Community,” NACE Rev. 115 and the “International Standard Classification of Occupations,” 1968,16 respectively. The work of the experts consisted of evaluating for each job the exposure to 70 agents, based on the general questionnaire, the specialized questionnaires (if administered), and their own experience in the field. Indices to assess for each exposure included the expert's confidence in the presence of the exposure (categorized as possible, probable or certain), the frequency of exposure defined as the percentage of working time exposed (categorized as 1–5, 5–30, or >30%), and the intensity of exposure (categorized as low, medium, or high). To obtain standardization in the application of the intensity index, cut-points between low, medium, and high intensity for each exposure were defined quantitatively where possible. For crystalline silica, less than 0.05 mg/m3 represented low exposure, 0.05–0.2 mg/m3 medium exposure, and more than 0.2 mg/m3 high exposure. All exposure assessments were conducted without awareness of case–control status.
The experts were instructed to assess exposure to crystalline silica in situations in which fine dust is produced by mechanical wear of sand and various stones or materials (during sand blasting, mining, quarrying) or when silica materials are heated to high temperature (knocking out of sand molds in foundries, calcining of diatomaceous earth). To ensure homogeneity, exposure to silica from sand, concrete dust, cement dust, and brick dust, not involving mechanical wear or high temperatures, was not directly assessed by the experts. Instead, silica exposure from these sources was calculated through algorithms, using the expert evaluation of exposure. For each given compound, levels of intensity were considered depending on the crystalline silica content (eg, exposure to concrete dust represented a higher level of exposure to free silica than exposure to cement dust). Where crystalline silica content was judged to be low, the level of frequency and confidence were lowered accordingly to reduce the final quantity of exposure. For example, if brick dust was assessed to be present with intensity (x), frequency (y), and confidence (z), the algorithm for exposure to crystalline silica was intensity (x − 1), frequency (y − 1), and confidence (z). These algorithms differed for sand, concrete, and cement dust, reflecting their silica content.
Standardization of the application of the exposure assessment methodology was reinforced through yearly workshops and coding exercises. The reliability of the experts’ assessments was evaluated through an interteam agreement study of 19 job descriptions concluding that the quality of the individual expert teams was comparable but that the expected levels of misclassification greatly differed between the agents assessed.17
We applied unconditional logistic regression modeling to study the association between silica and lung cancer. All models were adjusted for age (9 categories), center (15 categories), tobacco consumption (pack-years, continuous), and age at which full time education ended to represent socioeconomic status (4 categories), with categories based on distributions in controls. All other occupational agents assessed in this study were also considered as potential confounders for the association between silica and lung cancer by fitting dichotomous variables for these agents in the model consecutively. Results are presented as adjusted odds ratios (ORs) with 95% confidence intervals (CIs).
Various strategies were applied to calculate the lifetime exposure to silica. The first measure was “duration (years),” which was expressed as the total number of years the individual worked in an occupation in which exposure to silica occurred. The second measure, “weighted duration (hours),” represented the total number of hours effectively exposed, based on the total duration in years (each year counting for 2000 working hours) multiplied by the assessed frequency of exposure (0.03 for low frequency, 0.175 for medium frequency, 0.65 for high frequency). The third measure, “cumulative exposure,” represented the total number of hours a person was effectively exposed, multiplied by the intensity level assessed for each exposed year. Silica intensity levels (low, medium, and high) received weights in the analysis of 0.025 mg/m3, 0.1 mg/m3, and 0.5 mg/m3, respectively, reflecting the midpoints of the quantitative intensity scale set for crystalline silica. For categorical analyses, exposure categories were based on tertiles or quartiles of the exposure distribution in exposed controls; individuals who had never been exposed to respirable crystalline silica were used as the reference. The presence of a linear trend was studied by fitting the categorical variable for cumulative exposure as a continuous variable in the model. Typically, a latency period of at least 15 to 20 years between the exposure and the clinical manifestation of disease is observed. Therefore, all dose–response associations were studied after including a 20-year lag period, in which the 20 years before interview were considered not exposed.
Stratified analyses were performed by sex, country, age at interview, smoking status, and age finished full-time education, and heterogeneity in risk-estimates among these strata was assessed. Given the fact that increased risk for all main histologic types of lung cancer has been observed for other occupational risk factors such as asbestos, we took advantage of the study's large size to investigate the association between crystalline silica exposure and lung cancer histology. We also investigated associations resulting from exposure to primary construction materials (sand, cement, concrete, and brick dust) and from employment in specific industrial settings in which the prevalence of exposure is high but for which relatively few epidemiologic studies have been conducted. Separate analyses were conducted for lung cancer cases with concomitant silicosis and their associated controls. Population attributable risks were calculated based on the odds ratio and the fraction of exposed controls in the study population.18 Statistical analyses were performed with statistical analysis packages; SAS for Windows release 8.02 (SAS Institute, Cary, NC) and STATA release 7.0 (Stata Corporation, College Station, TX).
In total, 2852 cases and 3104 controls were eligible to participate in the study (Table 1). Overall, the response rate was 84% for cases and 85% for controls. Of the 542 (16%) eligible cases not included in the study, 27 had been discharged from hospital before the interview, 53 were too ill to be interviewed, 13 had died before the interview, and 449 refused to participate. Of the 552 (15%) eligible controls not included in the study, 16 had been discharged from hospital before the interview, 21 were too ill to be interviewed, 2 had died before interview, 511 refused to participate, and 2 were excluded from the analyses because their age was missing. Cases had on average 3.1 jobs for which a general questionnaire was completed whereas controls had 2.6 jobs. Both cases and controls had on average 0.86 specialized questionnaires. The quality of the interview was scored by the interviewer for cases (70% good, 23% medium, and 7% poor) and controls (78% good, 18% medium, and 4% poor). No differences between cases and controls were observed in either the interview quality or the number of questionnaires completed.
As expected, the proportion of smokers (including ex-smokers) was greater in cases (90%) compared with controls (64%). Age finished full-time education was used instead of a more direct measure of education due to difficulties in standardizing educational levels across 7 European countries. A higher proportion of cases spent less time in education with 63% leaving full-time education by their eighteenth birthday, compared with 56% of controls.
Lifetime prevalence of exposure to silica was 15% for cases and 10% for controls (Table 2). Models were adjusted consecutively for all assessed occupational agents. The odds ratio (OR) changed appreciably only after adjustment for wood dust and insulation dust (including asbestos). After adjustment for age, sex, center, education, insulation dust and wood dust, the OR of lung cancer for ever-silica exposure was 1.44 (95% CI = 1.22–1.69). Further adjustment for smoking reduced the OR to 1.37 (1.14–1.65). Sex-specific ORs were 1.32 (1.10–1.59) in men and 2.07 (0.91–4.74) in women. The population attributable risk associated with occupational exposure to crystalline silica for both sexes was 4%; the figure was 5% among men and 1% among women. There was country variability in lifetime prevalence of exposure to silica for cases (range, 8–23%) and controls (range, 6–17%), but we did not observe heterogeneity in risk estimates among countries (P = 0.21). Silica exposure did not appear to be a risk factor for cases diagnosed with lung cancer at a relatively young age (<55 years) but was a risk factor for lung cancer cases ages 55 to 65 years (OR = 2.10; CI = 1.54–2.85). Current smokers ever exposed to silica had a 41% increased lung cancer risk (1.41; 1.07–1.87) compared with unexposed current smokers. Similar increases in risk were also observed for ex-smokers (1.31; 0.99–1.73) and never smokers (1.41; 0.79–2.49), with no evidence of heterogeneity among these risk estimates (P = 0.37).
To assess the possible exposure–response relationship, adjusted ORs for lung cancer were calculated for cumulative exposure, duration, and weighted duration, based on a lag of 20 years (Table 3). Clear trends of increasing risk with increasing cumulative exposure were observed for cumulative silica exposure (1.07, 1.06, 1.47, and 2.08 at 0–9, 9–35, 35–200, and >200 mg/m3-hours, respectively; P for trend <0.0001), duration of exposure (OR = 1.25, 1.06, 1.44, and 1.73 at 0–2, 2–5, 6–14, and >14 years, respectively; P for trend = 0.0001) and weighted duration of exposure (OR = 1.16, 1.22, 1.31, and 1.88 at 0–350, 350–1200, 1200–4000, and >4000 hours, respectively; P for trend <0.0001).
With regard to lung cancer histology and occupational exposure to crystalline silica, the results indicate a clearly increased risk for the main histologic types of lung cancer, squamous cell carcinoma (OR = 1.45; CI = 1.15–1.83) and adenocarcinoma (1.49; 1.09–2.03; Table 4). The risk estimates were, however, highest for the categories with highest cumulative silica exposure for squamous-cell carcinoma, small-cell carcinoma and adenocarcinoma. Overall, small and large cell carcinomas showed lower risk estimates, although linear trends of increased OR with cumulative silica exposure were observed.
Analyses of the lung cancer risk associated with exposure to different sources of silica revealed clear linear trends of increasing ORs for silica directly assessed by experts (P for trend <0.0001), silica from sand (P for trend = 0.01), silica from concrete (P for trend = 0.03), silica from cement (P for trend = 0.02), and silica from any source material (P for trend <0.0001; Table 5). Increased risk of lung cancer was restricted to those groups with the highest cumulative exposure (>100 mg/m3-hours) to silica directly assessed by experts (2.18; 1.53–3.12), silica from concrete (2.11; 1.09–4.11) and silica from any source (1.95; 1.46–2.59).
We observed elevated risks of lung cancer for ever-exposure to silica in the construction (1.27; 1.01–1.60), manufacturing (2.03; 1.30–3.17) and mining (1.48; 1.02–2.13) industries (Table 6). In addition, clear linear trends of increasing ORs were observed for the construction (P for trend = 0.005), manufacturing (P for trend = 0.003) and mining (P for trend = 0.03) industries, even though the dose-response relationship was not always monotonic.
Crystalline silica repeatedly has been found to be associated with lung cancer in highly exposed industrial situations but has rarely been studied in a community-based case–control design because of limitations in exposure assessment or study size. Our case–control study of almost 6000 individuals investigated exposure–response relationships for lung cancer in relation to silica exposure, adjusted for smoking and occupational confounders. The results showed trends of increasing risk of lung cancer with increasing cumulative exposure to silica and with duration of exposure. Overall, these results are consistent with a causal interpretation of the association between exposure to silica and lung cancer, but no conclusions on the role of silicosis in this causal pathway can be drawn due to small numbers and lack of information on medical diagnosis of silicosis.
In contrast to Northern and Western European countries, relatively few studies on occupational exposure and risk of lung cancer have been published in Central and Eastern European countries. It has therefore been difficult to present an accurate picture of occupational exposure to carcinogens. Many of the Central and Eastern European countries participating in this study had large-scale surveillance programs, based on periodic x-ray examinations, which may be possibly an indicator of exposure to an occupational hazard such as crystalline silica.19 In general, it is difficult to obtain reliable quantitative exposure data in retrospective case–control studies and, in common with previous epidemiologic studies of silica and lung cancer, our study is limited due to the lack of detailed historical data. Our analyses were based on semiquantitative exposure data to elucidate the exposure–response relation and minimize confounding effects due to smoking and other potential occupational exposures. An obvious limitation of this study is the difficulty in validating the exposure estimates in the absence of a standard.20 Although no information was available on the actual crystalline silica exposure experienced by the study participants, we did have information on the specific tasks performed, the commencement of employment. and the duration of exposure. Exposures were estimated by local experts, using complete occupational histories. As well as exposure duration, experts assessed exposure intensity and confidence enabling exposure-response analyses.
Because of the retrospective nature of the method used in this study, exposure misclassification is of concern and may obscure associations. Exposure assessment was conducted in 2 stages: first, centrally trained interviewers performed a person-to-person interview, using structured questionnaires; then, local experts performed case-by-case exposure assessment, unaware of case-control status. Extensive training was undertaken throughout the study to ensure standardization of the exposure assessment process. Despite training, the inter-rater agreement for directly assessed silica was poor.17 However, the majority of silica exposure was indirectly assessed by applying algorithms to exposure estimates for sand, cement dust, concrete dust and brick dust. Excellent agreement was observed for sand (kappa = 0.86), cement dust (kappa = 0.82), and concrete dust (kappa = 0.75) and fair-to-good agreement was observed for brick dust (kappa = 0.62). The inter-rater agreement for the final silica exposure estimates, combining direct and indirect estimates, was 0.35. Because the exposure assessment was conducted without knowledge of case–control status, misclassification of exposure is expected to be nondifferential and, therefore, is most likely to result in attenuation of any existing exposure-response relationship. However, misclassification is unlikely to account for the positive exposure–response relationship observed here. If there were no true exposure–response trend, nondifferential (with respect to outcome) and random misclassification of exposure would be very unlikely to produce an apparent positive trend in either our categorical or continuous analyses.21–24 The population in this study was drawn from countries with medium and high incidence rates of lung cancer and included predominantly industrialized regions, and we did not observe substantial heterogeneity in crystalline silica risk estimates between countries. In addition, confounding by country-specific risk factors is an unlikely explanation for the observed associations in this study, given that we adjusted for center in the analyses.
The studies that provide the most convincing evidence of carcinogenicity indicate that increased risks of lung cancer are restricted to those groups with the highest cumulative exposure to crystalline silica.6 In a nested case–control study of North American industrial sand workers, Hughes et al25 reported that cumulative exposures of 1.8–4.5 mg-yr/m3 were associated with a relative risk of lung cancer of 2.15 compared with those with exposures <70 mg-yr/m3. Steenland and Sanderson26 reported an excess of lung cancer in an industrial sand workers cohort (standardized mortality ratio = 1.6; 95% CI = 1.31–1.93) and a trend of increasing risk with increasing cumulative exposure. A pooled analysis of 10 cohorts of silica-exposed workers for which quantitative data on exposure were available showed that the log of cumulative exposure lagged 15-years was an important predictor of lung cancer. In addition, categorical analyses by quintiles of cumulative exposure showed clear trends of increasing odds ratios.24 Results from these industry-based studies are particularly useful because there is relatively little potential for exposure to confounding occupational carcinogens.
Our results are consistent with previous cohort studies, which have shown up to a two-fold increase in lung cancer risk in those groups with the highest cumulative exposure to crystalline silica. However, our case-control approach offers certain advantages over industrial cohort studies, including allowing a detailed adjustment for smoking in the analyses. Although smoking was by far the largest risk for lung cancer, our results did not suggest an interaction between tobacco smoking and exposure to silica on the risk of lung cancer beyond a multiplicative model. The increased risk of exposure to crystalline silica was, however, less apparent after adjustment for smoking, indicating the possibility of some confounding from smoking. Previous population-based studies have found similar, elevated risks of lung cancer for individuals occupationally exposed to crystalline silica,27,28 although our finding of exposure–response relationships dependent on the source of the silica exposure, while employed in specific industrial settings, and for the main histologic types of lung cancer is new. Another advantage of the current study is its size, with detailed exposure data on 821 silica-exposed cases and controls. Women have been under-represented in silica-related research, either because previous studies did not have sufficient power or they were conducted in male-dominated industries. This is the first study, to our knowledge, that is sufficiently powerful to point to a moderately increased risk of lung cancer for women exposed to crystalline silica. Although country-specific ORs indicated elevated risks for exposure to silica, pooled exposure–response analyses appeared particularly helpful to strengthen the evidence for an association for otherwise relatively weak associations for exposure to silica. ORs for lung cancer showed a clear positive trend with increasing cumulative exposure. Potential occupational confounding exposures asbestos and polycyclic aromatic hydrocarbons were considered, but they did not appreciably modify the OR. In addition, the results did not change substantially when individuals with silicosis were excluded from the analysis.
In this case–control study, lung cancer risk can be studied in relation to exposure levels that are common in a wide variety of occupations and industries and can thus give an indication of the lung cancer risk attributable to occupational exposure to silica in the study population. Our results show that silica is associated with an increased lung cancer risk, independent of the source of the silica. An increased risk of lung cancer was observed in the mining, manufacturing, and construction industries, with a clear dose–response relationship in the latter. Although our results reflect risk for exposure to crystalline silica that is primarily quartz, it cannot be assumed that the cristobalite or amorphous silica were absent. Indeed, in manufacturing environments that involve heating processes there may be exposure to 2 or more polymorphs of crystalline silica, specifically to quartz or cristobalite. However, cancer risks from occupational exposure cannot yet be attributed to a particular polymorph because differences in their carcinogenic potential have not been established.1 There has been a significant decline in silica exposure levels over the years, although certain industries and occupations still experience elevated exposure levels to silica.13,29 A recent study in the U.S. construction industry provided evidence that a relatively large proportion of construction workers are overexposed to silica levels in excess of the current occupational exposure limit.12 Our results provide further evidence of the potential silica-related lung cancer burden in this global industry. Indeed, our results support the hypotheses that silica is an important risk factor for lung cancer, accounting for approximately 4% of lung cancers in our population, and further supports the IARC evaluation that crystalline silica is a human carcinogen.
The evidence of a carcinogenic effect of silica dust has been challenged, and the demonstration of a positive exposure-response can provide convincing evidence in favor of causality.3 The results derived from this multicenter case–control study indicate the positive relationship between silica exposure and lung cancer and provide further evidence for a causal association between silica and lung cancer. Our results suggest a positive trend of increasing risk of lung cancer with increasing cumulative exposure and duration of exposure to silica, with a 2.1-fold risk for those in the highest cumulative exposure quartile. This risk could not be explained by exposure to other occupational carcinogens or smoking, and was present for different sources of silica exposure, for different industrial settings and for the main histologic types of lung cancer.
1. International Agency for Research on Cancer. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans: Silica and Some Silicates
. Lyon, France: IARC; 1997.
2. Wong O. The epidemiology of silica, silicosis and lung cancer: some recent findings and future challenges. Ann Epidemiol
3. Steenland K. One agent, many diseases: exposure-response data and comparative risks of different outcomes following silica exposure. Am J Ind Med
4. Kurihara N, Wada O. Silicosis and smoking strongly increase lung cancer risk in silica-exposed workers. Ind Health
5. Hessel PA, Gamble JF, Gee JB, et al. Silica, silicosis, and lung cancer: a response to a recent working group report. J Occup Environ Med
6. Health & Safety Executive. Respirable Crystalline Silica—Phase 2: Carcinogenicity. Hazard assessment document. EH75/5
. London: HSE Books; 2003.
7. Parks CG, Cooper GS, Nylander-French LA, et al. Comparing questionnaire-based methods to assess occupational silica exposure. Epidemiology
8. de Klerk NH, Musk AW. Silica, compensated silicosis, and lung cancer in Western Australian goldminers. Occup Environ Med
9. Attfield MD, Costello J. Quantitative exposure-response for silica dust and lung cancer in Vermont granite workers. Am J Ind Med
10. Cherry NM, Burgess GL, Turner S, et al. Crystalline silica and risk of lung cancer in the potteries. Occup Environ Med
11. Checkoway H, Hughes JM, Weill H, et al. Crystalline silica exposure, radiological silicosis, and lung cancer mortality in diatomaceous earth industry workers. Thorax
12. Rappaport SM, Goldberg M, Susi P, et al. Excessive exposure to silica in the US construction industry. Ann Occup Hyg
13. Yassin A, Yebesi F, Tingle R. Occupational exposure to crystalline silica dust in the United States, 1988–2003. Environ Health Perspect
14. Siemiatycki J, Nadon L, Lakhani R, et al. Exposure assessment: In: Siemiatycki J (Ed). Risk Factors for Cancer in the Workplace
. Boca Raton, FL: CRC Press; 1991:45–115.
15. European Economic Commission. Statistical classification of economic activity in the European community. NACE Rev 1, 1990.
16. International Labour Office. International Classification of Occupations
. 2ed ed. Geneva: International Labour Office Publications; 1968.
17. ′t Mannetje A, Fevotte J, Fletcher T, et al. Assessing exposure misclassification by expert assessment in multicenter occupational studies. Epidemiology
18. Bruzzi P, Green SB, Byar DP, et al. Estimating the population attributable risk for multiple risk factors using case-control data. Am J Epidemiol
19. Boffetta P, Mannetje A, Zaridze D, et al. Occupational X-ray examinations and lung cancer risk. Int J Cancer
20. ′t Mannetje A, Steenland K, Attfield M, et al. Exposure-response analysis and risk assessment for silica and silicosis mortality in a pooled analysis of six cohorts. Occup Environ Med.
21. Flegal KM, Keyl PM, Nieto FJ. Differential misclassification arising from nondifferential errors in exposure measurement. Am J Epidemiol
22. Correa-Villasenor A, Stewart WF, Franco-Marina F, et al. Bias from nondifferential misclassification in case-control studies with three exposure levels. Epidemiology
23. Steenland K, Deddens JA, Zhao S. Biases in estimating the effect of cumulative exposure in log-linear models when estimated exposure levels are assigned. Scand J Work Environ Health
24. Steenland K, Mannetje A, Boffetta P, et al. International Agency for Research on Cancer. Pooled exposure-response analyses and risk assessment for lung cancer in 10 cohorts of silica-exposed workers: an IARC multicentre study. Cancer Causes Control
25. Hughes JM, Weill H, Rando RJ, et al. Cohort mortality study of North American industrial sand workers. II. Case-referent analysis of lung cancer and silicosis deaths. Ann Occup Hyg
26. Steenland K, Sanderson W. Lung cancer among industrial sand workers exposed to crystalline silica. Am J Epidemiol
27. Siemiatycki J, Gerin M, Dewar R, et al. Silica and cancer associations from a multicancer occupational exposure case-referent study. IARC Sci Publ.
28. Bruske-Hohlfeld I, Mohner M, Pohlabeln H, et al. Occupational lung cancer risk for men in Germany: results from a pooled case-control study. Am J Epidemiol
29. Stewart AP, Rice C. A source of exposure data for occupational epidemiology studies. Appl Occup Environ Hyg
© 2007 Lippincott Williams & Wilkins, Inc.