Article Text

Download PDFPDF

Possible effect of environmental exposure to asbestos on geographical variation in mesothelioma rates
  1. Stephen Goldberg1,4,
  2. Grégoire Rey2,
  3. Danièle Luce3,
  4. Anabelle Gilg Soit Ilg4,
  5. Patrick Rolland5,
  6. Patrick Brochard6,
  7. Ellen Imbernon4,
  8. Marcel Goldberg3
  1. 1INCa, Boulogne-Billancourt, France
  2. 2CépiDc, INSERM, Le Vésinet, France
  3. 3U 687, INSERM-Université Versailles Saint Quentin, Villejuif, France
  4. 4Département Santé Travail, InVS, Saint-Maurice, France
  5. 5Département Santé Travail, InVS, Bordeaux, France
  6. 6Laboratoire Santé, Travail, Environnement, Bordeaux, France
  1. Correspondence to Stephen Goldberg, Département Santé Travail, Institut de Veille Sanitaire, 12, rue du Val d'Osne, Saint-Maurice Cedex 94415, France; s.goldberg{at}


Background In population-based mesothelioma studies in industrialised countries, the incidence of mesothelioma without any identified asbestos exposure (IAE) is usually higher among women, while male incidence is mainly attributed to IAE. Through a comparison of the spatial distribution of male and female rates, and IAE and no IAE incidence, this study investigated whether mesotheliomas without IAE are in fact induced by non-recognised asbestos exposure, mostly from environmental sources.

Methods We calculated mesothelioma mortality (SMR) and incidence (SIR) ratios by district in France, pooling 30 and 10 years of data, respectively. Using correlation coefficients, we compared geographical patterns of male and female mesothelioma ratios, and IAE and no IAE mesothelioma ratios.

Results The raw numbers of male and female mesothelioma cases were equivalent. Mesothelioma SMR (0.76) and SIR (0.80) geographical correlations between men and women were strongly positive. SIR correlation between occupationally IAE and no IAE cases was also positive (0.69). Correlation between occupationally IAE and no IAE cases was positive among women but not among men.

Conclusions Data analyses of mesothelioma mortality and incidence showed that female cases occur in the same geographical areas as male cases. Female mesotheliomas with no IAE occur in the same geographical areas as exposed cases, suggesting asbestos has a major influence on female mesothelioma, likely through environmental exposure.

  • Mesothelioma
  • asbestos
  • environmental exposure
  • geographical distribution
  • France
  • Epidemiology
  • Statistics

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Asbestos is a recognised human carcinogen. It is causally related to mesothelioma of the pleura and of other sites, to lung cancer,1 and was recently recognised by the International Agency for Research on Cancer (IARC) as inducing laryngeal and ovarian cancer.2 Associations with other cancer sites, notably the colon, have been suggested but have not been definitively established. Asbestos also causes various non-cancerous respiratory disorders such as asbestosis and pleural plaques.3

In industrialised countries, asbestos-related occupations are the major sources of asbestos exposure, for instance mining and crushing, shipbuilding, manufacturing, the construction and demolition of buildings, lagging, insulation, elimination of waste, and the servicing and maintenance of asbestos-containing products and buildings. Apart from occupational sources, people can also be exposed to asbestos in other situations. Domestic exposure occurs notably among people living with asbestos workers. Environmental exposure may result from natural asbestos in the soil, living near asbestos mines or manufacturing plants, and living or working in asbestos-insulated buildings.4 It is likely that the vast majority of asbestos-induced disease in industrialised countries is caused by occupational rather than non-occupational asbestos exposure,1 2 5 a legacy of the frequent past use of asbestos. Concern used to be focused on the occupational environment, but it is now recognised that asbestos fibres are widely distributed in the general environment. Indeed, exposure to asbestos in occupational settings has decreased sharply, but this reduction is probably less for non-occupational exposure.6 7 While exposure in non-occupational locations is generally much lower than in occupational settings, levels are still significant and are spread over a larger section of the population.4

It is therefore important to assess the burden of disease induced by environmental exposure to asbestos. Mesothelioma is the best model for that purpose, since it is considered to be a specific outcome of asbestos exposure, no causal factors except for exposure to asbestos (and some other mineral fibres) having been established or suspected.5 Moreover, it rarely occurs spontaneously; the background incidence is estimated to be of the order of one case per million population per year.8

There are some methodological difficulties when studying the effects of environmental exposure to asbestos. In industrialised countries and in asbestos-producing regions of developing countries, the high prevalence of occupational exposure tends to mask non-occupational exposures, thus making it difficult to isolate the effects of the latter. Moreover, while lifelong occupational exposure is relatively easy to assess through structured interviews in population-based studies, environmental exposure is frequently unknown by the subjects and only very scarce data are available on individual lifelong cumulative exposure in non-occupational settings. Thus, when considering the potential impact of environmental exposure to asbestos on the total number of mesothelioma cases in a population, the available data do not allow direct estimations.

An alternative approach is to study the association of mesothelioma spatial distribution between IAE (identified asbestos exposure) and non-IAE cases, assuming that IAE cases are a proxy of asbestos exposure. Consequently, a positive and strong association between non-IAE and IAE cases would suggest that at least some of the non-IAE cases are caused by asbestos exposure. The latter approach is followed in this study.

Two types of available data can be used with this method: mesothelioma mortality data available for the entire French territory and covering many years, for which exposure status is unknown, and mesothelioma incidence data, available for a fraction of the territory for a 10-year period, for which asbestos exposure has been assessed.

Under some additional assumptions, there are two reasons why IAE and non-IAE case distribution can be compared when examining male and female mesothelioma mortality. First, studies of environmental exposure to naturally occurring asbestos in the soil in different regions of the world, where the history of exposure among women and men is similar, have reported mesothelioma sex ratios (men:women) close to one4; accordingly, the risk of developing a mesothelioma induced by asbestos does not differ by gender. Second, in population-based mesothelioma studies in industrialised countries, the proportion of cases without IAE is usually around 10–20% among men but much higher among women (50–60%), depending on the population under study and the methods of exposure assessment.9 10

Provided the spatial heterogeneity of female mesothelioma mortality is large enough to be at least partially attributable to non-IAE, if male and female mesothelioma mortalities are spatially associated, any other important mesothelioma risk factor would have to have the same geographical distribution as asbestos, and its temporal evolution would have to be the same as the use of asbestos to explain the parallel rise in male and female mesothelioma incidence,6 all of which seems unlikely.

Quantification of the association between non-IAE and IAE mesothelioma incidence is more direct but has a lower power. Therefore, analyses of both data sources should provide complementary information.

The aim of this work is to test the hypothesis that the role of asbestos exposure in mesothelioma incidence is underestimated by successively comparing the geographical patterns of male and female mesothelioma mortality and of IAE and non-IAE incidence rates.

Population and methods


We used two data sets: mortality data provided by the French Death Register (INSERM-CépiDc) and incidence data from the French National Mesothelioma Surveillance Program (PNSM). Since 1998 the PNSM has recorded incident pleural tumours in 26 French geographical districts or ‘départments’ (out of 96 in metropolitan France), covering a population of approximately 16 million people, a quarter of the French population. In 2006, the population of these 26 districts ranged from 129 000 to 1 916 494, while the population of all French metropolitan districts ranged from 77 500 to 2 583 493. The PNSM allows the use of a standardised procedure of pathological and clinical diagnosis confirmation, and lifetime exposure to asbestos from different sources is reconstructed.10

Mortality data for metropolitan France were available for 1974–2005. This period covers three versions of the International Classification of Diseases (ICD): ICD-8 (1974–1978), ICD-9 (1979–1999) and ICD-10 (2000–2005). We selected 26 074 deaths certificates (30.1% women) coded as malignant neoplasm of the pleura or pleural mesothelioma (table 1). Sex, age, year of death and district of residence at the time of death were also recorded. Standardised mortality ratios (SMR) were computed by sex and district, using annual population estimates provided by the Institut National de la Statistique et des Études Économiques (INSEE).

Table 1

Numbers of mesothelioma deaths* according to period and ICD version

Incidence data were collected through the PNSM. All subjects with suspected mesothelioma living in participating districts were investigated and confirmed. From 1 January 1998 to 1 December 2008, 1937 (21.2% women) mesothelioma patients were recorded. Sex, age, year of diagnosis and district of residence at the time of diagnosis were registered. Occupational and domestic asbestos exposure history was assessed by trained interviewers using a structured questionnaire further coded, together with job titles, by experienced industrial and environmental hygienists for 1228 confirmed mesothelioma patients (the other 709 had missing exposure values, mainly due to incomplete interviews of the next of kin of deceased patients). Domestic exposure includes not only clear domestic exposure (eg, the use of asbestos containing devices at home) but also do-it-yourself building work and para-occupational exposure (eg, washing asbestos contaminated clothes at home). Exposure from industrial sources in the vicinity of the residence or from naturally occurring asbestos in the soil, that is environmental asbestos exposure, was not assessed.

For each potential source, reported asbestos exposure was classified as not identified, possible, probable or very probable. Finally, Boolean exposure variables were constructed separately for occupational and domestic exposure: exposed (very probable or probable) or non-exposed (possible or not identified). In order to avoid the powerful confounding factor of occupational exposure to asbestos and to isolate the effect of domestic exposure, the latter is only considered among cases without identified occupational exposure, cases with very probable, probable or possible occupational exposure being excluded.

The combination of these two exposure variables was named ‘combined exposure’. If the individual was found to have been exposed to asbestos, whether occupationally or domestically, then his or her combined status was ‘exposed’. If the individual was not found to be exposed in either setting, the combined variable value was ‘not exposed’.

Standardised incidence ratios (SIR) were computed by sex, asbestos exposure status and district, using annual population estimates provided by the INSEE in districts covered by the PNSM.

Statistical analysis

SMRs were computed for the entire mainland French territory by district, separately for each sex. The entire French population was the reference population.

SIRs were computed for the 26 districts covered by the PNSM by district, separately for each sex. SIRs for women and men separately were also computed according to occupational and non-occupational exposure status. The entire French population was the reference population. Correlation of SIRs was analysed by sex (men versus women) without regard for exposure status, and then by exposure status (exposed versus non-exposed). SIR correlation by exposure status was also analysed separately within each sex. SIRs of occupationally exposed men were also correlated with those of non-exposed women.

The strength of the linear relationship between SMRs and SIRs was estimated by Pearson correlation coefficients (r). 95% CIs were computed for Pearson correlation coefficients with a Monte Carlo routine (10 000 samples).

In order to minimise the influence of underpopulated districts, the calculation of correlation coefficients was weighted on the sum of mesothelioma expected deaths or expected cases in the district.



Male and female mesothelioma mortality rates at the level of the district can be compared visually using the map in figure 1 which is based on data from 1974 to 2005. Rates are spatially very heterogeneous over the whole country and higher among men (0.84–5.08 per 100 000 men) than among women (0.11–1.62 per 100 000 women). As regards geographical patterns, districts with rates higher than the national average correspond to heavily industrialised areas, to coastal districts with shipyards and to specific large asbestos processing plants in rural settings.

Figure 1

Mesothelioma standardised mortality rates in French districts (1974–2005).

The mesothelioma mortality rates of men and women were also compared using SMRs at scale of the district (table 2). Correlation for each ICD period was tested separately: all correlation coefficients r were positive, being over 0.5. When all periods were pooled, r was increased to 0.76 (95% CI 0.31 to 0.84).

Table 2

Mortality geographical patterns: correlations between male and female SMRs


Mesothelioma cases recorded by the PNSM were pooled over a decade and distributed according to their exposure to asbestos (table 3). About a third of mesothelioma cases (709) were not interviewed, mostly because they had died before the survey took place. Of the remainder, over 81% of men had been occupationally exposed, but only 22% of women. Among cases with no identified occupational exposure, domestic exposure was recorded for approximately 17% of men and women, without gender specificity. The vast majority of mesothelioma cases with asbestos exposure, whether occupational or domestic, are men.

Table 3

Distribution of incident cases among women and men according to asbestos exposure status*; (1998–2008)

Overall, 26.5% of mesotheliomas do not have IAE (326 not exposed out of 1228 investigated). It must be noted that the number of cases with no IAE is almost the same in men (154) as in women (172), which is striking since the mesothelioma standardised incidence rate for cases with IAE is 0.83 per 100 000 men, almost ten times the corresponding 0.09 per 100 000 women.

Analyses of correlation of the geographical patterns are presented in table 4. The SIRs of men and women were compared at the scale of the district, regardless of exposure, yielding a correlation coefficient r of 0.80 (95% CI 0.49 to 0.87). Among non-exposed subjects, the SIRs of men and women were correlated (r=0.57), and even more strongly correlated among those exposed to asbestos (r=0.81). When both sexes were pooled together, the SIRs of cases with occupational exposure were correlated with unexposed cases (r=0.69, 95% CI 0.14 to 0.84), which was not the case for domestic exposure, while combined exposure gave a correlation coefficient of slightly under 0.5 (r=0.45), both with inconclusive CIs. Among men, no geographical correlation was found between cases with and without occupational exposure, but among women geographical correlation was 0.59 for occupational exposure, although with a large CI. In addition, the SIRs of occupationally exposed men compared with non-exposed women were found not to be correlated (r=0.12, 95% CI −0.47 to 0.41).

Table 4

Incidence geographical patterns: correlations between SIRs


Pooling mortality data over three decades and three ICD versions in order to produce mesothelioma mortality rate gender maps is justified by the necessity of having enough death certificates to allow comparison between genders at the district level. The latency between exposure to asbestos and the occurrence of mesothelioma is more than 30 years, with death almost certain a short time later. So the period of exposure we examined spans the period from the end of World War II to the mid-1970s, when the first regulation designed to reduce asbestos exposure was enforced in France. This period, roughly 1945–1975, was when maximum asbestos exposure, mainly occupational, occurred in France.11 The fact that spatial distribution showed a similar pattern of high geographical disparity (except for shipbuilding districts along the English Channel, Atlantic and Mediterranean coasts where asbestos was intensively used in industries operated by a mostly male workforce) in men and women suggests asbestos was the common cause of mesothelioma.

The long latency period probably means that an unknown proportion of fatal or incident cases moved away from the district where they had been exposed to asbestos before they were diagnosed. This dilution of the geographical link between asbestos exposure and mesothelioma thus limits the power of our geographical approach.

Nevertheless, correlations of geographical patterns of male and female mesothelioma cases were consistent whether computed from mortality or incidence data. This means that, even though the fraction of female mesothelioma cases attributable to occupational asbestos exposure is small (38.5% estimated in France through PNSM data compared to 83.2% for men),10 females cases occur in the same areas as male cases.

Incidence data, either of both sexes pooled together or only for women, showed that mesothelioma cases with no IAE occur in the same geographical areas as occupationally exposed cases. Since it was precisely among women that the geographical distribution of non-IAE cases matched that of IAE cases, this suggests a major influence of asbestos.

Consequently, a large number of non-IAE cases are likely to have been in fact exposed to asbestos in the environment. This environmental exposure is a combination of geographically diffuse exposure (for instance through construction of buildings with asbestos-containing materials) and of specifically located sources of exposures (shipyards, asbestos-processing factories), which are difficult to separate since precise life-long environmental exposure is almost impossible to assess.

Among men, non-IAE cases seemed to be geographically distributed somewhat independently of IAE cases, while non-IAE cases among women are also distributed independently of IAE male cases. This could be explained by the large percentage of occupational asbestos exposure-related male mesotheliomas related to workers in confined areas or small workplaces, such as electricians or plumbers working in construction,12 all geographically widely spread without yielding significant environmental pollution, whereas male and female mesotheliomas induced by environmental exposure are more likely to arise from proximity to large asbestos-processing factories located in specific places.13–23 This is further suggested by the similarity in the number of male and female cases with no IAE among incident mesothelioma cases, moreover geographically correlated, suggesting that environmental asbestos exposure due to proximity to asbestos-related industries equally affects both genders.

The coherence of these results seems to rule out the hypothesis of a mesothelioma risk factor other than asbestos specific to women.

Population-based case–control studies that integrate the effects of different exposure circumstances experienced by a given population might thus be used to assess the proportion of mesotheliomas due to environmental exposure to asbestos. In a study carried out in three European countries, no evidence of occupational exposure to asbestos was found for 53 out of 215 cases, yielding 24.6% non-occupational cases17; in a case–control study nested within the PNSM,10 the percentage of non-occupational cases was 18% (84 out of 466 cases). Among the 198 mesotheliomas with a histological diagnosis for which exposure was defined in the Italian National Mesothelioma Register, 73 (37%) had no occupational exposure.24

Also a proportion of mesothelioma have no identified exposure to asbestos by any route (occupational or non-occupational). In the European study, among the 41 subjects with no evidence of occupational exposure to asbestos with sufficient exposure data, nine (22%) were classified as having no exposure to domestic or environmental sources,17 while the rate for the Italian National Mesothelioma Register was 29.5%.24 Our study shows a similar figure of 26.5%.

This suggests that there is a real burden of environmental asbestos exposure in industrialised countries that could account for approximately 20% of all mesotheliomas. However, extra research is needed if these findings are to be generalised, leading to a better estimation of the attributable fraction of the mesotheliomas caused by environmental exposure to asbestos.


We showed that in France female and male mesothelioma cases are located in the same geographical areas. Moreover, the fact that female mesotheliomas with no IAE occur in the same geographical areas as exposed cases, strongly suggests a major influence of asbestos on female mesothelioma, likely through environmental exposure. The high proportion of female mesothelioma cases with no IAE also suggests that the burden of environmental asbestos exposure in industrialised countries is far from negligible.

What this paper adds

  • In population-based studies in industrialised countries, the fraction of mesothelioma incidence with no identified asbestos exposure (IAE) is usually higher among women, while male incidence is mainly attributed to occupationally IAE.

  • Geographical analyses have shown that female mesothelioma cases without IAE occur in the same geographical areas as male asbestos exposed cases, suggesting a major influence of asbestos in female mesothelioma, likely through environmental exposure.

  • The findings suggest that the burden of environmental exposure to asbestos can be estimated from mesothelioma cases without IAE.


Data used for this work were kindly provided by the Centre d'épidémiologie sur les causes médicales de Décès (INSERM-CépiDc), and by the Programme National de Surveillance du Mésothéliome (PNSM). We also thank Charlotte Roudier-Daval from the Institut National du Cancer (INCa) for making the maps.



  • Funding This study was funded by the Institut de Veille Sanitaire (InVS), 12 rue du Val d'Osne, 94415 Saint-Maurice Cedex, France and the Institut National du Cancer (INCa), 52 avenue André Morizet, 92513 Boulogne-Billancourt Cedex, France.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.