Characteristics of the seven studies which fit the inclusion criteria
| Authors, year of publication, industry, location | Study duration and follow-up since first exposure | Exposure measurement | Diagnostic classification | Silicosis cases/sample size; risk | Analysis methods | Mean silicosis latency (years) | Mean intensity (mg/m3) | Loss to follow-up/ censoring |
| Cross-sectional studies | ||||||||
| Hnizdo and Sluis-Cremer, 1993 Gold mines; South Africa19 | Employed ≥10 years 1968–1971. Subsequent annual radiograph; post-employment follow-up to mean 1982 (range 1968–1991) via ‘occasional radiological examination’. Mean 24 years service (range 10–43) | Airborne shift dust sampling from 20 gold mines and 13 locations in mines by thermal precipitator and konimeter. Approximate silica content 30%24 37 | ILO. Earliest radiograph ≥1/1* Blind reading by three trained readers; one who correlated best with autopsy chosen | 313/2235 (14%) | Cumulative risk – SAS PROC LIFETEST (Life table) Cumulative risk model – SAS PROC LIFEREG; Loglogistic | 34.6 | 0.087 | 0/2235 (0%) |
| Steenland and Brown, 1995 Gold mines; USA14 | Employed ≥1 year underground 1940–1965. Follow-up to 1990 via death certificate and radiographs during two cross-sectional surveys in 1960 and 1976. Average follow-up 37 years, average employment 9 years | 1153 personal air samples with cyclone-filter. Silica via infrared spectrometry38 | ILO. Earliest radiograph ≥1/1 or ‘small opacities’ and death certificate. Unclear radiology method | 170/3330 (5%) 128 cases by death certificate only, 29 by x-ray only, 13 by both | Cumulative risk – specified formula† | Not reported | 0.09 14% no data on exposure after 1975; possible underestimation | 67/3330 (2%) |
| Hughes, et al
1998 Diatomaceous mine and processing; USA22 | Employed >1 year 1942–1987. ‘Periodic radiographs’. Median employment 5.5 years (range 1–49.3 years). Median follow-up 29.9 years (range 1–53 years)39 | 5174 airborne shift samples from particle impinger and gravimetric pump. Silica content (10–25%) estimated based on materials processed39 40 | ILO. Earliest radiograph ≥1/0 or large opacities. At least 2 of 3 certified ‘B’ readers for +ve result | 81/1809 (4.5%) | Cumulative risk – SAS PROC LIFETEST (Life table) Cumulative risk model – SAS PROC LIFEREG; Loglogistic | 11.4 | 1452/1809 had mean exposure <0.5 mg/m3
357/1809 had mean exposure >0.5 mg/m3 | Unclear. 214/2342 (9%) ‘unknown’ at study end in 199439 |
| Chen et al, 2005 Tin mines, tungsten mines, pottery factories; China10 | Started work after 1950, worked ≥1 year 1960–1974. Yearly radiographs since 1963. Follow-up to 1994. Minimum follow-up 26 years. Mean years of dust exposure pottery 24.9 years, tin 16.4 years, tungsten 16.5 years | Airborne gravimetric sampling from historical samples. Dust and silica content validated by side-by-side gravimetric sampling and X-ray diffraction | Chinese classification system. Earliest radiograph stage I or above. Agreement of 2 of 3 radiologists required | 2816/14 427 (20%) tungsten miners, 855/4028 (21%) tin miners, 785/4547 (17%) pottery workers | Cumulative risk – SAS PROC LIFETEST (Life table) Cumulative risk model – SAS PROC LIFEREG; Weibull | Tungsten: 19.0 Tin: 20.2 Pottery: 29.4 | Tungsten: 0.2 Tin: 0.152 Pottery: 0.254 | Tungsten 82/9007 (0.9%) Tin: 12/4028 (3%) Pottery: 4/4547 (0.1%) |
| Liu et al, 2013 Metal mines, pottery factories; China25 | Worked ≥1 year 1960–1974. All participants followed up to 2003 unless died or lost to follow-up. Yearly radiographs since 1963. Average follow-up 34.5 years | Airborne gravimetric sampling from historical samples. Dust and silica content validated by side-by-side gravimetric sampling and X-ray diffraction | Chinese classification system. Radiographs stage I or above. No radiology method reported | 5297/34 018 (16%):
| No silicosis analytical methods | Not reported | Tungsten: 0.52 Iron: 0.08 Pottery: 0.15–0.30 (1960–1980), 0.12–0.15 (post 1990) | 1527/34 018 (4.5%) |
| Cross-sectional studies | ||||||||
| Kreiss and Zhen, 1996 Hardrock mine; USA18 | Random sample of dust exposed, aged >40 years in mining town. Radiographs performed in 1984 or 1986. Mean time from first exposure 36.1 years, minimum 13 years | Airborne samples; 649 job title-specific gravimetric dust measurements and 484 silica measurements between 1974 and 1982 | ILO. Median of three B readers; radiologic profusion of >1/0 | 32/100 (32%) | No cumulative risk. Modelling: logistic regression | Not reported | Silicotics: 0.08 Non-silicotics: 0.06 | 6/100 (6%) miners no exposure data |
| Miller et al, 1998 Coal mine; Scotland11 | Available radiographs from any years of 1970/1974/1978. Outcome based on follow-up radiograph in 1991. Minimum follow-up 13 years. Pit closure 1981 | Regular (<yearly) airborne samples, initially thermal impinger then gravimetric pump. Silica via infrared spectrometry41 | ILO. Most recent radiograph ≥2/1. Median result of three ‘experienced readers’ | 47/547 (9%) | No cumulative risk. Modelling: logistic regression | Not reported | Not reported | NA. However study is of 547/1416 (39%) traceable survivors |
| Rosenman, et al 1996 Foundry; USA23 | Employed ≥5 years before 1986; medical records up to 1991 used. Most recent radiographs obtained via work records or ‘other means’. Average follow-up 28.3 years and employment 19.2 years | Airborne samples; midget impinger. Silica conversion via bulk samples % | ILO. Most recent radiograph ≥1/0. At least 2 of 3 certified ‘B’ readers for +ve result | 36/936 (4%) | No silicosis analytical methods | Not reported | 0.23 | NA |
A valid correspondence between Chinese stage I, II, III and ILO 1, 2, 3 exists.42
*Autopsy performed in 131 miners with confirmation of silicosis in 128; 3 false positives.
†Cumulative risk (H(t)) = 1− exp[−sum of (hazards*interval width)]. All studies used job exposure matrices for personal exposure measurements. Study conversion factors were used where necessary.
ILO, International Labour Organisation.