Open Access

Research Article

Materials and Methods Top

The factory and the town

The AC plant in Casale Monferrato had been active from 1907 to 1985, producing plane and corrugated sheets, pipes and pressure pipes, and special AC products. The size of the work force varied over time and never exceeded 1,500 workers. In 1981 the company reported using 15,000 tons of asbestos (10% crocidolite) (Magnani et al. 1996). Processes gave origin mostly to diffuse emissions from building openings such as shed windows. Local exhausts were not installed until 1978 and were limited to the lathes used for machining pressure pipes. Only the power plant had a stack, which entailed no emission of asbestos fibers (AF). The factory is upwind from the town—about 1,500 m from the center and 250 m from the closest residential areas. As far as we can infer from its raw materials and products, airborne emissions from the AC plant included both chrysotile and crocidolite fibers, and the same should apply to waste-waters and waste materials. Environmental asbestos concentration was measured shortly before the factory shutdown (1985) or afterward but never during the 1950s and 1960s when the plant was fully active and when, considering mesothelioma latency, the most relevant exposures presumably occurred. Estimates reported here are the average of repeated measurements and, if not otherwise specified, they refer to airborne AF with length > 5 μm and diameter > 0.3 μm.

Marconi et al. (1989) reported asbestos concentrations ranging from 11 AF/L close to the plant (around 400 m), to 4.5 AF/L in the city center and to 1 AF/L in the city area farthest away in 1985; at that time production had already been significantly reduced. Short-term (4–8 hr) air sampling was employed. Scanning electron microscopy (SEM) was used for fiber counting (detection limit: 0.4 F/L) and energy dispersive X-ray analysis (EDXA) for fiber identification, according to the Asbestos International Association (AIA) method (AIA 1984).

The Local Health Authority (LHA) measured annual average concentrations < 1 AF/L in 1990–1991, with 12% of samples exceeding 1 AF/L (Unità Sanitaria Locale di Casale Monferrato, unpublished data). Short-term (4–6 hr) air samples were taken. Fiber counts (detection limit, 0.3 F/L) were conducted by SEM, and fibers were identified as asbestos fibers by EDXA. Sampling and analytical AIA methods were followed, but only fibers with diameter > 0.5 μm were counted.

In 1991 asbestos fibers concentrations from 2.2 to 7.4 AF/L were reported in the residential areas of Casale. The average concentration of total (any length) asbestos fibers was 48.4 AF/L (1.5 AF/L total amphiboles), versus, respectively, 0.2–12.1 and 0.0–0.2 AF/L in other industrial cities (Chiappino et al. 1991, 1993). Very long sampling times (3–7 days) were used. Fibers were counted on transmission electron microscope (TEM, detection limit not provided) and were identified by EDXA. Fibers were suspended in solvent after ashing the original cellulose acetate filter (or a portion of it), with ultrasound agitation and eventually refiltered on a nucleo-pore membrane for “indirect” TEM preparation. This technique clearly differs from the AIA method. Marconi et al. (1989) stated that calcium-rich amphiboles accounted for 15–30% of all asbestos fibers. Chiappino et al. (1991) reported that amphiboles represented almost 50% of asbestos fibers > 5 μm. In the LHA report no asbestos type–specific figure was given.

A survey on lung fiber burden in a series of consecutive necropsies supports the hypothesis that important pollution from amphiboles was present in Casale (Magnani et al. 1998). This notion is obviously relevant for our study, as the carcinogenic potency of amphiboles for inducing MM is considered much higher than that of chrysotile (Hodgson and Darnton 2000).

Study design

This is a population-based case–control study that includes cases of pleural MM newly diagnosed between 1 January 1987 and 30 June 1993 among residents in Casale Monferrato and the surrounding area, comprising roughly 52 towns and over 100,000 inhabitants (of whom 40,000 are in Casale).

Cases were retrospectively identified through surveys of the pathology units of the hospitals serving the study area and were all histologically confirmed. Of the 123 cases included in the Piedmont Registry of Malignant Mesotheliomas files for 1987–1993 among residents in the study area, 116 (94.3%) were eligible for this study (Ivaldi et al. 1999). Controls were selected randomly either from the files of residents in the LHA or from the mortality files of residents in the same area and individually matched to cases by sex, birth date (± 18 month), vital status, and date of death (± 6 months). In subsequent analyses, individual matching was disregraded because in our study, matching variables were spatially neutral (Cuzick and Edwards 1990; Diggle 2003).

Live subjects and the closest relative of deceased subjects were interviewed from 1993 to 1995 using a standardized questionnaire, with sections designed to reconstruct the lifelong occupational history of the subjects, their spouses, relatives and any other cohabitants, and others on demographic characteristics, smoking, radiation treatment, and schools attended.

In addition to environmental exposure, represented by the distance between residence and the factory, three other main sources of asbestos exposure were identified: a) occupational exposure in the AC industry; b) domestic exposure, with which we refer to either the indoor presence of asbestos materials such as asbestos fabrics of ironing tables, fire-proof sheets for stoves and ovens, or AC materials and roofings in very close proximity to the house (e.g., garden, courtyard); and c) occupation in the AC industry of relatives and cohabitants. These variables were coded as dicotomic (yes/no) for all subjects. Occupational exposure in the AC industry was chosen as a proxy to asbestos occupational exposure tout court because it corresponds to high intensity of exposure and is highly specific. In addition to AC production and related activities (warehousing and transportation of raw asbestos and final products), no other noticeable source of asbestos exposure of industrial origin was recorded in Casale (Magnani et al. 1991). AC products were used in Casale Monferrato as elsewhere in the building industry, namely, for roofs and water pipings. There is anecdotical evidence of small quantities of asbestos having been used in a factory producing printing machines in the gaskets for the hoods of the drying sections and in a textile (silk) workshop in pipe laggings. Such exposure circumstances were actually less common than in most industrial settings in our study region. Therefore, confounding due to residual occupational exposure is unlikely. More details on data collection procedures and exposure coding have been published elsewhere (Magnani et al. 2001).

Because the focus of our present study was on the risk associated with residential distance from AC plant, special care was dedicated to the questionnaire’s section designed to reconstruct the complete residential history of all subjects, comprising all the addresses of subjects (within and outside Casale), and a description of each dwelling and its neighborhood environment. To have a comprehensive estimate of asbestos domestic exposure, we collected information on the presence and use of asbestos materials in the house or its proximities. All residential addresses obtained from the original questionnaires were compared with and completed by information from the town office registries, and coded as Universal Transverse Mercator (UTM) geographic coordinates using a global positioning system (GPS) receiver. The geographic coordinates of the AC factory location were determined in a similar manner. Because each subject had inhabited more than one dwelling, the address of the longest-held residence was chosen as a proxy to residential distance exposure, after exclusion of dwellings occupied in the last 20 years before the date of diagnosis for cases or before the date of the interview or the date of death for alive and deceased controls, respectively.

Informed consent was obtained before the study from live subjects or from the next-of-kin of deceased subjects. The study was approved by the institutional review board of the department of Biomedical Sciences and Human Oncology of the University of Turin.

Statistical methods

Basic data analyses used logistic regression (Breslow and Day 1980) and provided risk estimates in form of ORs, adjusted for age and sex, with 95% confidence intervals (CIs).

Kernel density estimation was used to estimate and map spatial variation in disease risk (Bithell 1990; Kelsall and Diggle 1995). It was assumed that cases and controls formed two independent point-Poisson processes of different intensities. Kernel smoothing was used to estimate the density surfaces of cases and controls separately, and the risk surface was then obtained as the ratio of the two.

To analyze specifically the effects of residential distance from the AC plant on MM risk, we first defined a logistic regression model in which distance was subdivided in classes and included as a categorical variable. Furthermore, we defined a mixed additive-multiplicative model in which the odds of disease are as follows:

where w is a proportionality factor; γ_j is the log OR for the jth multiplicative risk factor, z_j ; α₁ is the excess risk for AC occupational exposure (AC); f(·) is the distance excess risk function; and d = x–x₀ is the distance (meters) between individual residence location and the source. The distance excess risk function was specified as:

known as an exponential-threshold model (Waller and Pocquette 1999), where the parameter α₂ represents the excess relative risk at the source and the parameter β models the exponential risk decay (per unit squared-distances). Multiplicative risk factors included age, sex, indoor or outdoor domestic exposure to asbestos, and occupation in the AC industry of any relative. An additive scale was assumed for the risk associated with residential distance and AC occupational exposure. The rationale of this choice is to ensure that the risk is unchanged at and infinite distance from the source x₀. Furthermore, it is reasonable, on biological grounds, to assume that different exposures to the same agent interact additively (Pira et al. 2005).

Spatial clustering was investigated through nonparametric methods including the Cuzick–Edwards test (Cuzick and Edwards 1990) and second-order distance methods based on Ripley’s K and Diggle’s D functions (Diggle and Chetwynd 1991; Ripley 1976).

Results Top

Of 116 cases and 330 controls eligible for the study, 103 (89%) and 272 (82%) agreed to participate in the study and were interviewed. There were no significant differences in age, sex, or residence between respondents and nonrespondents.

Table 1 describes the complete data set of 375 interviewed subjects. Cases and controls presented similar distributions by age, sex and vital status. Age and sex were nevertheless always included in the logistic models as potential confounders.

Geographic coordinates of the longest-held residence of each subject up to 20 years before the date of diagnosis (cases) or of the interview/death (controls) were available for 97 (94%) cases and 250 (92%) controls. The distribution of index residences in a geographic area of roughly 50 km² around Casale is shown in Figure 1. Seventeen controls were outside the plot area. Longest-held addresses could be considerable distances from Casale, as subjects were only required to reside in the LHA of Casale at the moment of diagnosis. The spatial distribution of the controls on the map represents the population density in the region.

Figure 2 shows the contour plot of the risk for MM in the same geographic area. The risk surface was estimated as the ratio of case to control kernel density surfaces, including all individuals independent of their status of occupational exposure. Risk shows a well-defined peak in the center of Casale, southeast from the AC factory, and seems to decrease monotonically in all directions.

The effects on MM risk of the three main sources of asbestos exposure that are different from residential distance exposure are shown in Table 2. Twenty-eight cases and 14 controls had worked in the AC industry, resulting in an OR of 7.1 (95% CI, 3.5–14.3) for occupational exposure. Between domestic exposure and occupational exposure of relatives, the latter was the most important, although risk estimate decreased (maintaining statistical significance) after adjusting for AC occupation.

To analyze the residential distance exposure effects, we classified individuals as resident in geographic bands at increasing distance from the AC plant (Tables 3 and 4). Table 3 shows that risk decreases with increasing distance, with strong evidence of a spatial trend (p < 0.0001). The band at 3–5 km from the AC factory included a remarkably high concentration of cases. Restricting the analysis to the nonoccupationally exposed subjects (Table 4) did not change risk estimates significantly. The robustness of risk estimates associated with residential distance allowed us to include all subjects in the subsequent analyses rather than restricting them to the nonoccupationally exposed subgroup, thereby improving the precision of the estimates.

Table 5 shows the relative risk for MM associated to AC occupation, domestic exposure, and relatives’ AC occupation, as estimated by the mixed additive–multiplicative model for excess relative risk, with and without adjustment for the distance of the residence from the AC plant.

Estimates of the parameters of the mixed additive-multiplicative model (Equation 1) were α₂ = 9.5 (95% CI, 2.8–49.1), which represents residential distance excess risk at the source, namely, a relative risk of 10.5 for hypothetical residents living at the AC plant (at zero distance from the source); and β = 0.11 × 10⁻⁷ (95% CI, 0.48 × 10⁻⁸ to 0.25 × 10⁻⁷), which represents risk decay rate per unit squared-distances (distance measured in meters) moving away from the source. At 10 km from the source, the relative risk was estimated to have decreased by about 60%, from 10.5 to 4.2 (still remarkably high). Figure 3 shows the expected relative risk by distance obtained from the exponential decay with threshold model and the point estimates of Table 3. The estimated effect of AC occupational exposures on MM risk is approximately 3 times that of residential distance exposure at the source.

Spatial clustering of cases was statistically significant according to the T ₃ Cuzick-Edwards test (p = 0.003). Second-order distance methods showed evidence of spatial aggregations of cases of average dimension approximately 300 m (Diggle’s D-test p = 0.016). Similar results were obtained both when the analyses were restricted to non-occupationally exposed subjects and when all subjects were included.

Discussion Top

This study has shown substantial effects of environmental asbestos exposure on mesothelioma risk in an Italian town with an important asbestos cement plant. Available data on asbestos fibers concentration in Casale had already suggested relatively high levels of exposure, with average concentration of 48.4 AF/L (total fibers) and of 1.5 AF/L (total amphiboles) in 1991 (Chiappino et al. 1991; Marconi et al. 1989). These measurements, however, are inadequate to describe fiber concentration at a small area scale. Therefore, we have estimated an exposure–response relationship using the distance from the AC factory as a proxy for environmental asbestos exposure. Two key sets of findings are discussed below.

First, there was a substantial risk from residential distance exposure. Residence at the location of the asbestos cement factory had a relative risk for mesothelioma of 10.5 (95% CI, 3.8–50.1), adjusted for occupational and domestic exposures. Risk decreased rapidly with increasing distance from the factory, but at 10-km distance the risk was still 60% of its value at the source. Second, the estimate of AC occupational risk increased remarkably when residential distance exposure was considered, showing that the relative risk from occupational exposures had been underestimated. The relative risk for occupational exposure was 6.0 (95% CI, 2.9–13.0), but this increased to 27.5 (95% CI, 7.8–153.4) when adjusted for residential distance exposure. The reason for such an increase is that when residential distance exposure is not accounted for, AC occupational risk is estimated by comparing occupationally exposed subjects with a reference group still being exposed to high levels of environmental asbestos, although not through occupation. Residential distance exposure was a very strong confounder of occupational exposure in our data.

Domestic and occupational exposures of any relative were comparatively less important risk factors. They were included in the models as potential confounders (Table 5).

The major strengths of this study are the completeness and reliability of information on residential histories of subjects. Since lifelong residential histories of deceased subjects (the majority) were reconstructed by interviewing relatives and could therefore be imprecise, all information was checked against and supplemented by records of town office registries. All addresses were then geocoded blindly with respect to the case–control status of the subject using a GPS. To account for mesothelioma latency, the addresses of the last 20 years before diagnosis for cases or before the interview/ death for controls were disregarded. We examined a number of different criteria for selecting individuals’ residences for spatial analysis, each of which could introduce a different bias. Among all addresses of each subject, we considered the closest to the factory (Magnani et al. 2001), the longest-held, the average coordinates of all dwellings weighted by the duration of residence, and others. We opted for the most specific longest-held residences.

Similarly, the choice of AC occupational exposure as a proxy for all occupational exposures favored specificity over sensitivity. However, in the area under study, other potential sources of occupational exposure to asbestos were scarce (Magnani et al. 1991) and not associated with the distance from the AC plant, causing at worst nondifferential misclassification.

Potential biases that could have affected the study concerned diagnostic criteria reliability, blindness in assessing both case and exposure status, and proportion of nonresponders; the manner in which these were controlled in the design of the study is described elsewhere (Magnani et al. 2001). A limitation of this study is the small numerosity, which is partially compensated by the richness of information on occupational and residential individual histories that are sufficient to provide robust risk estimates.

The choice of a mixed additive–multiplicative model to estimate the excess risk due to environmental exposure was suggested by the a priori consideration that such a model ensure that the excess relative risk estimate had the desirable property of tending to 1 at an infinite distance from the source. Including occupational exposure as well as residential distance as an additive term seemed plausible, as both exposures involve the same substance. A posteriori, the goodness of fit of the selected model (Equation 1) was better than that of a similar model in which occupational exposure was included as a multiplicative term, as shown by Akaike Information Criterion (AIC; Akaike 1974) of 361 and 365, respectively.

The functional form of risk dependency on the distance from the AC plant (exponential decay with a threshold) was chosen because this model provided the best fit to data. The AIC of a model including a simple exponential decay with the distance was higher. The inclusion of directional effects did not improve model fit. Other functional forms (risk peaked at a given distance from the source) could be considered (Figure 3), but the added complexity was not justified in terms of improved goodness of fit. However, there is still residual variability to be explained. In a previous analysis on residents in Casale, Magnani et al. (2001) observed that residential distance exposure risk remained high even at a considerable distance from the factory. This prompted the hypothesis that sources of exposure other than air pollution from the AC factory could be involved in the increased mesothelioma risk in the area. Among these sources is transportation of raw asbestos and AC materials to and from the railway station and to and from a warehouse. Furthermore, potential sources included the use of AC residuals such as mixing them into the soil to create hard pavement and improve water absorption or applying layers of finely ground AC—rich in asbestos fibers—in the lofts for thermal insulation. The effects of some of these sources were partially considered by adjustment in the models for domestic exposure, which comprised information on the presence of asbestos at home or in the courtyard or garden. Indeed, the results of spatial clustering tests give some support to the hypothesis of secondary sources of asbestos in the area. Further assessments are needed to understand their exact nature, location, and the entity of their effect, although this present study has clearly shown the major role of the AC factory as the principal source of asbestos pollution in the area of Casale.

In conclusion, this study provides strong evidence of an increased mesothelioma risk from asbestos environmental pollution from an industrial source. Although it is generally believed that environmental exposure is a less powerful determinant of MM risk than occupational exposure, this study shows that the strength of the effect of pollution from industrial sources on the general population can reach alarming magnitudes, up to one-third of the risk for asbestos industry workers. Conversely, failing to adjust for environmental exposure may lead to serious underestimation of the relative risk due to occupational exposure in population-based case–control studies carried on in such settings.

Figures and Tables Top

Figure 1.

Spatial distribution of the residences of both occupationally and nonoccupationally exposed cases and controls in a geographic area of approximately 50 km² around Casale. Residences are the longest-held among all residences of each individual after excluding 20 years before the date of the diagnosis of the index case. All 97 cases and 233 of 250 controls are included (17 control residences are outside the figure area). The location of the AC plant in the town of Casale is also indicated.

Figure 2.

Contour plot of the ratio of case to control kernel density surfaces including both occupationally and nonoccupationally exposed individuals. Smoothing parameters were 10 km for cases and 20 km for controls. The town of Casale and the location of the AC plant are indicated.

Figure 3.

Risk of malignant mesothelioma of the pleura in Casale in relation to the distance of individuals’ longest-held residence (after exclusion of 20 years from the date of the diagnosis) from the AC plant. Risk estimates are adjusted for age, sex, occupation in the AC industry, indoor or outdoor domestic exposure to asbestos (asbestos materials in the garden, courtyard, roof, inside the house), occupation in the AC industry of any relative. ORs are estimated through the logistic model and are represented with error bars (95% CI); relative risks are estimated through the model with exponential decay with threshold of the risk by distance from the AC plant and are represented as a smooth line.

Table 1.

Characteristics of study participants.

Table 2.

Risk of malignant mesothelioma of the pleura in Casale in relation to occupation in the AC industry, indoor or outdoor domestic exposure to asbestos (asbestos materials in the garden, courtyard, roof, inside the house), and occupation in the AC industry of any relative.

Table 3.

Risk of malignant mesothelioma of the pleura in Casale in relation to the distance of each individual’s longest-held residence (after exclusion of 20 years from the date of the diagnosis) from the AC plant.

Table 4.

Risk of malignant mesothelioma of the pleura in Casale for nonoccupationally exposed residents in relation to the distance of their longest-held residence (after exclusion of 20 years from the date of the diagnosis) from the AC plant.

Table 5.

Relative risk (RR) of malignant mesothelioma of the pleura in Casale in relation to occupation in the AC industry, domestic exposure, and occupation in the asbestos cement industry of any relative.

References Top

1. AIA 1984. Method for the Determination of Airborne Asbestos Fibres and other Inorganic Fibres by scanning Electron Microscopy. AIA Health and Safety Publ RTM no 2. London: Asbestos International Association.
2. Akaike H 1974. A new look at statistical model identification IEEE Transact Automatic Control 19: 716–722. Find this article online
3. Baris I, Simonato L, Artvinli M, Pooley F, Saracci R, Skidmore J, et al. 1987. Epidemiological and environmental evidence of the health effects of exposure to erionite fibres: a four-year study in the Cappadocian region of Turkey Int J Cancer 39: 10–17. Find this article online
4. Bithell JF 1990. An application of density estimation to geographical epidemiology Stat Med 9: 691–701. Find this article online
5. Bourdes V, Boffetta P, Pisani P 2000. Environmental exposure to asbestos and risk of pleural mesothelioma: review and meta-analysis Eur J Epidemiol 16: 411–417. Find this article online
6. Breslow NE, Day NE 1980. Statistical Methods in Cancer Research. Vol I: The Analysis of Case-Control Studies IARC Sci Publ 32: 5–338. Find this article online
7. Chiappino G, Sebastien P, Todaro A 1991. Atmospheric asbestos pollution in the urban environment: Milan, Casale Monferrato, Brescia, Ancona, Bologna and Florence [in Italian] Med Lav 82: 424–438. Find this article online
8. Chiappino G, Todaro A, Blanchard O 1993. Atmospheric asbestos pollution in the urban environment: Rome, Orbassano and a control locality (II) [in Italian] Med Lav 84: 187–192. Find this article online
9. Cuzick J, Edwards R 1990. Spatial clustering for inhomogeneous populations J Roy Stat Soc A 157: 433–440. Find this article online
10. Diggle PJ 2003. Statistical Analysis of Spatial Point Patterns. London: Arnold Publishers.
11. Diggle PJ, Chetwynd AG 1991. Second-order analysis of spatial clustering for inhomogeneous populations Biometrics 47: 1155–1163. Find this article online
12. Enterline PE 1983. Cancer produced by nonoccupational asbestos exposure in the United States J Air Pollut Control Assoc 33: 318–322. Find this article online
13. Gardner MJ, Saracci R 1989. Effects on health of non-occupational exposure to airborne mineral fibres IARC Sci Publ 90: 375–397. Find this article online
14. Goldberg M, Luce D 2005. Can exposure to very low levels of asbestos induce pleural mesothelioma? Am J Respir Crit Care Med 172: 939–940. Find this article online
15. Hammond EC, Garfinkel L, Selikoff IJ, Nicholson WJ 1979. Mortality experience of residents in the neighborhood of an asbestos factory Ann NY Acad Sci 330: 417–422. Find this article online
16. Hansen J, de Klerk NH, Musk AW, Hobbs MS 1998. Environmental exposure to crocidolite and mesothelioma: exposure-response relationships Am J Respir Crit Care Med 157: 69–75. Find this article online
17. Hillerdal G 1999. Mesothelioma: cases associated with non-occupational and low dose exposures Occup Environ Med 56: 505–513. Find this article online
18. Hodgson JT, Darnton A 2000. The quantitative risks of mesothelioma and lung cancer in relation to asbestos exposure Ann Occup Hyg 44: 565–601. Find this article online
19. International Agency for Research on Cancer 1987. Overall Evaluations of Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42 IARC Monogr Eval Carcinog Risks Hum Suppl 7: 1–440. Find this article online
20. Ivaldi C, Dalmasso P, Nesti M, Magnani C 1999. Malignant Mesothelioma Registry from Piedmont. Incidence in 1990–1995 [in Italian] Epidemiol Prev 23: 308–315. Find this article online
21. Kelsall JE, Diggle PJ 1995. Non-parametric estimation of spatial variation in relative risk Stat Med 14: 2335–2342. Find this article online
22. Luce D, Bugel I, Goldberg P, Goldberg M, Salomon C, Billon-Galland MA, et al. 2000. Environmental exposure to tremolite and respiratory cancer in New Caledonia: a case-control study Am J Epidemiol 151: 259–265. Find this article online
23. Magnani C, Agudo A, Gonzalez CA, Andrion A, Calleja A, Chellini E, et al. 2000. Multicentric study on malignant pleural mesothelioma and non-occupational exposure to asbestos Br J Cancer 83: 104–111. Find this article online
24. Magnani C, Borgo G, Betta GP, Botta M, Ivaldi C, Mollo F, et al. 1991. Mesothelioma and non-occupational environmental exposure to asbestos Lancet 338: 50. Find this article online
25. Magnani C, Dalmasso P, Biggeri A, Ivaldi C, Mirabelli D, Terracini B 2001. Increased risk of malignant mesothelioma of the pleura after residential or domestic exposure to asbestos: a case-control study in Casale Monferrato, Italy Environ Health Perspect 109: 915–919. Find this article online
26. Magnani C, Mollo F, Paoletti L, Bellis D, Bernardi P, Betta P, et al. 1998. Asbestos lung burden and asbestosis after occupational and environmental exposure in an asbestos cement manufacturing area: a necropsy study Occup Environ Med 55: 840–846. Find this article online
27. Magnani C, Terracini B, Ivaldi C, Botta M, Budel P, Mancini A, et al. 1993. A cohort study on mortality among wives of workers in the asbestos cement industry in Casale Monferrato, Italy Br J Ind Med 50: 779–784. Find this article online
28. Magnani C, Terracini B, Ivaldi C, Botta M, Mancini A, Andrion A 1995. Pleural malignant mesothelioma and non-occupational exposure to asbestos in Casale Monferrato, Italy Occup Environ Med 52: 362–367. Find this article online
29. Magnani C, Terracini B, Ivaldi C, Mancini A, Botta M 1996. Tumor mortality and from other causes in asbestos cement workers at the Casale Monferrato plant [in Italian] Med Lav 87: 133–146. Find this article online
30. Marconi A, Cecchetti G, Barbieri M 1989. Airborne mineral fibre concentrations in an urban area near an asbestos-cement plant IARC Sci Publ 90: 336–346. Find this article online
31. McDonald AD, McDonald JC 1980. Malignant mesothelioma in North America Cancer 46: 1650–1656. Find this article online
32. McDonald JC, McDonald AD 1996. The epidemiology of mesothelioma in historical context Eur Respir J 9: 1932–1942. Find this article online
33. Mirabelli D, Cadum E 2002. Mortality among patients with pleural and peritoneal tumors in Alta Valle di Susa [in Italian] Epidemiol Prev 26: 284–286. Find this article online
34. Newhouse ML, Thompson H 1993. Mesothelioma of pleura and peritoneum following exposure to asbestos in the London area. 1965 Br J Ind Med 50: 769–778. Find this article online
35. Pan XL, Day HW, Wang W, Beckett LA, Schenker MB 2005. Residential proximity to naturally occurring asbestos and mesothelioma risk in California Am J Respir Crit Care Med 172: 1019–1025. Find this article online
36. Paoletti L, Batisti D, Bruno C, Di Paola M, Gianfagna A, Mastrantonio M, et al. 2000. Unusually high incidence of malignant pleural mesothelioma in a town of eastern Sicily: an epidemiological and environmental study Arch Environ Health 55: 392–398. Find this article online
37. Pira E, Pelucchi C, Buffoni L, Palmas A, Turbiglio M, Negri E, et al. 2005. Cancer mortality in a cohort of asbestos textile workers Br J Cancer 92: 580–586. Find this article online
38. Ripley BD 1976. The second-order analysis of stationary point processes J Appl Probab 13: 255–266. Find this article online
39. Sakellariou K, Malamou-Mitsi V, Haritou A, Koumpaniou C, Stachouli C, Dimoliatis ID, et al. 1996. Malignant pleural mesothelioma from nonoccupational asbestos exposure in Metsovo (north-west Greece): slow end of an epidemic? Eur Respir J 9: 1206–1210. Find this article online
40. Schneider J, Rodelsperger K, Pohlabeln H, Woitowitz HJ 1996. Environmental and indoor air exposure to asbestos fiber dust as a risk and causal factor of diffuse malignant pleural mesothelioma [in German] Zentralbl Hyg Umweltmed 199: 1–23. Find this article online
41. Teta MJ, Lewinsohn HC, Meigs JW, Vidone RA, Mowad LZ, Flannery JT 1983. Mesothelioma in Connecticut, 1955–1977. Occupational and geographic associations J Occup Med 25: 749–756. Find this article online
42. Waller L, Pocquette CA 1999. The power of focused score tests under misspecified cluster models In: Lawson AB, Böhning D, Lesaffre E, Biggeri A, Viel J-F, Bertollini R, editors. Disease Mapping and Risk Assessment for Public Health. London: John Wiley & Sons. pp. 257–269.