Article Text

Download PDFPDF

Original article
Environmental asbestos exposure in childhood and risk of mesothelioma later in life: a long-term follow-up register-based cohort study
  1. Sofie Bünemann Dalsgaard1,2,
  2. Else Toft Würtz2,
  3. Johnni Hansen3,
  4. Oluf Dimitri Røe4,5,
  5. Øyvind Omland1,2
  1. 1 Department of Clinical Medicine, Aalborg University, Aalborg, Denmark
  2. 2 Department of Occupational and Environmental Medicine, Danish Ramazzini Center, Aalborg University Hospital, Aalborg, Denmark
  3. 3 Institute of Cancer Epidemiology, Danish Cancer Society, Copenhagen, Denmark
  4. 4 Department of Oncology, Aalborg University Hospital, Aalborg, Denmark
  5. 5 Department of Clinical and Molecular Medicine, Norwegian University of Science and Technology, Trondheim, Norway
  1. Correspondence to Sofie Bünemann Dalsgaard, Department of Occupational and Environmental Medicine, Danish Ramazzini Center, Aalborg University Hospital, Aalborg, Denmark; ksbunemann{at}


Objective To examine the risk of malignant mesothelioma (MM) in former pupils who attended primary school near an asbestos cement plant.

Methods A cohort of 12 111 former pupils, born 1940–1970, was established from individual historical records from four primary schools located at a distance of 100–750 m in the prevailing wind direction from an asbestos cement plant operating from 1928 to 1984 in Aalborg, Denmark. The school cohort and a comparison cohort consisting of 108 987 gender and 5-year frequency-matched subjects were followed up (2015) for MM in the Danish Cancer Registry. Using Cox regression, HRs were estimated for the incidence of MM. Adjustments for occupational and familial asbestos exposure were made with a job exposure matrix. An SIR analysis including latency periods testing the cancer incidence rate was performed with the comparison cohort as the reference rate.

Results The median person-years of follow-up were 62.5 years in the school cohort and 62.2 years in the comparison cohort. There were 32 males and 6 females of the former pupils who developed MM during the follow-up: HRmale 7.01 (95% CI 4.24 to 11.57), HRfemale 7.43 (95% CI 2.50 to 22.13). Those who attended school 250 m north of the plant had the highest HR for MM, 10.65 (95% Cl 5.82 to 19.48). No significant trend between school distance and risk of MM was established (p=0.35).

Conclusion Our results suggest that boys and girls who attended schools and lived in the neighbourhood of an asbestos cement plant later in life have a significantly increased risk of MM.

  • cancer
  • asbestos
  • environmental exposure
  • malignant mesothelioma
  • childhood

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.

Key messages

 What is already known about this subject?

  • It is well documented that asbestos exposure causes malignant mesothelioma.

  • Previous studies on malignant mesothelioma have mainly focused on occupational asbestos exposure in adulthood.

What are the new findings?

  • We found a significantly increased risk of malignant mesothelioma in a cohort that had been environmentally (predominantly chrysotile) asbestos-exposed during childhood.

How might this impact on policy or clinical practice in the foreseeable future?

  • Our study will have an impact on the compensation policy debate.

  • Today, in Denmark, compensation is only paid to mesothelioma cases who themselves have been occupationally asbestos-exposed or if a family member has been occupationally asbestos-exposed.


Malignant mesothelioma (MM) is a rare malignancy arising in the pleura, peritoneum, pericardium and in the tunica vaginalis.1 The association between occupational asbestos exposure and MM was first documented by Wagner et al. In the Wagner et al study, environmental exposure as a risk factor for MM was also shown, and several studies have verified that living nearby an asbestos-emitting source can increase the risk of MM.2 3 Marinaccio et al documented in a large epidemiological national surveillance study from Italy that 10.2% of MM cases are due to non-occupational exposure to asbestos.4 Magnani et al found an increased risk of pleural MM from environmental asbestos exposure, controlled for other sources of asbestos exposure, and suggested that environmental exposure caused a greater risk than domestic exposure.5 In the cities of Casale Monferrato and Bari in Italy, the mesothelioma risk increased with proximity of residence to an asbestos plant.6 7 To distinguish between occupational and environmental asbestos exposure, the Italian studies retrieved data by interview supplemented with register data. However, the attributable risks from Italy are place-specific and time-specific and cannot be extrapolated to account for the conditions in Denmark or other countries.

Denmark has, during the last century, been a producer of asbestos-containing products. In Denmark, nearly 90% of the imported raw asbestos was used in the manufacture of asbestos cement products. The only asbestos cement plant in Denmark was located in Aalborg in the North Denmark Region, a city reaching 100 000 inhabitants in the 1960s, and it was situated in a densely populated area of the city.8 From 1928, when the production began, and until 1984, a total of approximately 620 000 tons of asbestos were consumed (89% chrysotile).9

Only few studies have investigated whether asbestos exposure in childhood increases the risk of MM.10

The aim of our study is to analyse the impact of neighbourhood environmental asbestos exposure in childhood and the risk of mesothelioma later in life.

Materials and methods

Study design

This is a cohort study using individual data from high-quality national registers: the Civil Registration System (CRS),11 the Danish Cancer Registry (DCR)12 and the Danish Supplementary Pension Fund Register (ATP).13 Further, a country-specific job exposure matrix (JEM) was applied to assess occupational asbestos exposures.14


The four schools included were located at a distance of 100–750 m in the prevailing wind direction15 from the asbestos cement plant (see figure 1). Compulsory individual seventh-grade school records were used to identify former pupils. The pupils were identified by their name and birthplace or personal identification number, a unique 10-digit number all residents in Denmark are assigned from the CRS since April 1968. The flow chart (see figure 2) shows how the cohort was established. School records were excluded if either the personal identification number was missing or not validated in the CRS, or if a pupil was born before 1940 or after 1970.

Figure 1

The distribution of the four schools in an area of 250 m north and 100–750 m northeast of the asbestos plant. The picture shows the location of the asbestos plant. The arrow shows the prevailing wind direction from the plant.

Figure 2

Flow chart showing the establishment of the cohorts. CRS, Civil Registration System. 

From the CRS we obtained information on place of birth, civil status, emigration, disappearance, death, and identity of parents, siblings spouses and children.11 The unique identification number enables linkage of information between all national registers. The CRS was used to obtain a random cohort for comparison that was frequency matched (1:9) according to age and gender.

Subjects from both the school cohort and the comparison cohort were excluded because of emigration, death or cancer diagnosed before school start in the year of their 12th birthday. The final cohort comprised 12 111 former pupils (50.3% males and 49.7% females) and 108 987 comparison subjects.

Malignant mesothelioma

From the DCR we identified the MM cases registered in the period from April 1968 (start of CRS) to the end of 2015, including tumour characteristics. The DCR is regarded as virtually complete and contains cancer diagnosis registered since 1943, classified according to an extended Danish version of the International Classification of Diseases, 7th Revision (ICD-7), which included the diagnosis of mesothelioma (1943–1977), ICD-O (1978–2003) and ICD-10 (2004 and onwards).12 The MM diagnosis in the cancer registry covered validated ICD-10 codes for mesothelioma with location in the pleura, peritoneum and pericardium.

Assessment of asbestos exposure

High levels of airborne asbestos were measured at the asbestos plant and the relatively low-tech industry rendered workers highly exposed,16 and until 1972 asbestos was transported from the port to the plant in leaky hessian bags, with spill in the air.17 In this respect, we assume there has been substantial asbestos pollution from the plant to the neighbourhood.

We assume the environmental asbestos exposure for the school cohort began when the subjects became residents in the area close to the asbestos plant. From the CRS registry no exact residential history was available before 197718; therefore, data from seventh-grade school records have been used as a proxy for all childhood environmental asbestos exposure.

A Danish version of the Nordic Occupational Cancer Study (NOCCA) JEM was used to assess occupational asbestos exposure. The development of the NOCCA JEM has been described by Kauppinen et al; in brief, the occupational exposure was characterised by three metrics: the proportion of exposed subjects within a job, the mean level of exposure and four periods of calendar time (1945–1959, 1960–1974, 1975–1984 and 1985–1994).14 We have evaluated, edited and supplemented the NOCCA JEM for it to be compatible with the Danish industry codes (DSE77). Since 1 April 1964 all employees in Denmark have been compulsory members of the ATP, and information on all employments, including start and end dates, on a company and industry type level (DSE classification) has been registered and kept for wage earners aged 16–66 years working minimum 9 hours/week.13 Furthermore, the occupationally related asbestos exposure has been evaluated by two specialists in occupational medicine identifying local asbestos risk companies, for example, work with recycling of hessian bags and work in the asbestos cement plant.

A subject was defined as ever occupationally asbestos-exposed if the exposure prevalence in the JEM exceeded 50% in at least one job in the period from April 1964 until 31 December 1994. By this categorisation, 9.99% of the comparison cohort were occupationally asbestos-exposed. This is in line with previous estimation that approximately 150 000 persons in Denmark have been occupationally exposed to asbestos until the asbestos ban in 1986,19 corresponding to approximately 10% of the working population.

Relatives, that is, mother, father, siblings under 18, spouses and children were identified by the unique identification number in the CRS. To assess relatives’ occupational asbestos exposure, we extracted the employment history from the ATP and used the edited NOCCA JEM. Relatives were defined as ever occupationally asbestos-exposed if exposure took place in the period from the index subject was born (the earliest April 1964) to the 18th birthday (the legal age of majority in Denmark). Spouses and children were defined as ever occupationally asbestos-exposed if the index subject was 18+ years old and under 18 years old, respectively. Relatives’ occupational asbestos exposure is in our study termed ‘familial occupational asbestos exposure’. An individual from the school cohort was defined as environmentally asbestos exposed in the absence of both occupational asbestos exposure and familial occupational asbestos exposure.


The χ2 test was used to compare categorical variables between the two groups of the study population. Age medians were calculated by the Wilcoxon-Mann-Whitney test. Associations between environmental asbestos exposure and mesothelioma were tested by regression analysis based on a Cox proportional hazards model with adjustment for occupational and familial asbestos exposure. The analyses were supplemented by test for trend using school distance as an ordinal variable. We used likelihood ratio test to examine the interactions between occupational and familial asbestos exposure.

Person-years at risk for each of the two cohorts were calculated from 2 April 1968 (start of CRS) or 1 August, the year the subjects turned 12 years old (seventh-grade school start) if born after 2 April 1968. Accumulation of person-years at risk ended at the date of diagnosis of MM, date of death, emigration or disappearance or on 31 December 2015, whichever came first.

Sensitivity analyses were performed by (1) recoding subjects without ATP data (self-employed) into having occupational asbestos exposure, (2) excluding all subjects born prior to year 1948 in order to have complete employment records, (3) excluding all subjects not born in Denmark, (4) excluding all subjects born after year 1955 to find the HR for the oldest age groups, and using different cut-points for categorising a subject as occupationally asbestos-exposed by exposure prevalence, (5) 0%, (6) 10%, (7) 25% and (8) 75%.

We estimated SIR as the ratio of the observed and the expected cases with 95% CI. Further, lag time analyses deferring start of follow-up (10, 20 and 30 years) were performed for the SIR (see online supplementary table s1).

Supplementary file 1

All analyses were performed using Stata V.15.1.


The main characteristics of the cohorts are described in table 1. In the school cohort the median follow-up time was 62.5 years and and for the comparison cohort was 62.2 years.

Table 1

Description of the study population

All subjects in the school cohort have been environmentally asbestos-exposed. Further potential occupational asbestos exposure was detected in 2048 subjects (16.9%) in the school cohort and in 10 888 subjects (10.0%) in the comparison cohort. Subjects with familial occupational asbestos exposure were estimated as 1916 subjects (15.8%) in the school cohort, and in the comparison cohort 10 398 subjects (9.5%) had relatives with asbestos-related work.

Mesothelioma risk

A total of 76 incident MM cases (38 MM cases in the school cohort and 38 MM cases in the comparison cohort) occurred during the follow-up period (table 2). This represented an elevated relative risk for the school cohort compared with the comparison group (SIR 8.77; 95% CI 6.38 to 12.05). The school cohort had a HR for MM of 7.15 (95% CI 4.54 to 11.27), adjusted for index subject’s occupational asbestos exposure and familial occupational asbestos exposure (table 3). The HR for males was HRmale 7.01 (95% CI 4.24 to 11.57); the unadjusted HR for females was 7.43 (95% CI 2.50 to 22.13).

Table 2

Description of malignant mesothelioma cases

Table 3

Risk of malignant mesothelioma in relation to school attended near an asbestos cement plant

No evidence of interaction between index subjects’ occupationally asbestos-exposure and relatives’ occupationally asbestos-exposure was found (results not shown). Restricting the analysis to subjects with neither occupational nor familial occupational asbestos exposure resulted in a HR for MM of 5.12 (95% CI 2.49 to 10.57). The distribution of cases regarding anatomical site was not significantly different in the two cohorts (table 2).

School distance and mesothelioma risk

Table 3 presents the analysis of distance between the asbestos cement plant and the school attended and risk of MM. The highest HR was found for children who attended school at about 250 m north of the plant: HR for MM 10.65 (95% CI 5.82 to 19.48). No significant trend between school distance and risk of MM was established (p=0.347).

Characteristics of asbestos exposure in mesothelioma cases

In the school cohort 29% (6 males/5 females, ratio 1.2:1) had no other known asbestos-assessed exposure than the environmental asbestos exposure from attending school and living nearby the asbestos plant. Out of 38 cases in the comparison cohort, 58% (15 males/7 females, ratio 2.1:1) had neither occupational nor familial occupational asbestos exposure.

Using our JEM, 61% (23 males) from the school cohort and 39% (15 males) from the comparison cohort worked in an industry with potential asbestos exposure with a mean cumulated time of employment of 7.1 years (range 0.1–35.1) and 8.3 years (range 0.1–42.2), respectively. Table 4 shows the details on occupational asbestos exposure among mesothelioma cases; shipyard workers were the most prevalent, followed by workers in carpentry, electrical and insulations firms, and in total three persons were employed at the asbestos cement plant. For the shipyard workers, nine out of the ten cases in the school cohort had worked in the Aalborg Shipyard, while all shipyard workers in the comparison cohort had worked in shipyards elsewhere.

Table 4

Work industries for mesothelioma cases with occupational asbestos exposure

Sensitivity analyses

In a sensitivity analysis 134 subjects from the school cohort and 4176 subjects from the comparison group without ATP records were added to be in an occupationally and/or familial occupationally asbestos-exposed group. This did not change the adjusted HR for MM significantly (see online supplementary table s1 [a]).

We performed a subanalysis excluding all subjects born before April 1948, that is, aged at least 16 years old, having available employment information on all subjects (see  online supplementary table s1 [b]). This did not change the HR either. All mesothelioma cases had information about their employment history from the ATP.

In order to exclude potential natural-occurring asbestos exposure in childhood before the seventh grade, a sensitivity analysis was made restricting the analysis to subjects born in Denmark; no change in HR was found (see online supplementary table s1 [c]). The HR did not change either for those born before year 1955 (see online supplementary table s1 [d]). The majority (89%) of all cases were born before 1955. The HRs calculated with the variant proportions of asbestos workers are also with overlapping CIs and considered statistically similar (see online supplementary table s1 [e–h]).

Age at diagnosis and latency time

No significant difference in the median age at diagnosis was found in the cohorts. In the school cohort the median age at diagnosis for those only environmentally asbestos-exposed was 61.5 years. For those also occupationally and/or familial asbestos-exposed, the median age at diagnosis was 60.9 and 61.6 years in the school cohort and comparison cohort, respectively. Lag time analyses deferring start of follow-up 10, 20 and 30 years (see online supplementary table s2) show that the majority of cases (97%) developed MM more than 30 years after their seventh-grade school attendance.


In this register-based cohort study, we found, as assumed for an environmentally asbestos-exposed cohort, a similar HR for MM among males and females. We found a significantly higher risk for MM for pupils who attended either of the four schools included in the study, but no trend between school distance to the asbestos plant and risk of MM could be established.

Further, the risk of MM in the school cohort was also significantly higher when the analysis was restricted to subjects without occupational or familial occupational asbestos exposure.

Our results are in line with previous studies finding non-occupational exposure to asbestos to be associated with increased MM risk.20 21 The SIR of 8.77 (95% CI 6.38 to 12.05) included all asbestos exposure groups and was found to be in the same range as the SIR calculated by Mensi et al from Broni, an Italian town which had an asbestos cement factory operating for more than 60 years (SIR 8.4; 95% CI 7.1 to 9.9).22

Our results on mesothelioma risk are also consistent with the results by Magnani et al, who found an increased risk of MM for people who never worked in the asbestos cement industry and who attended their grammar school in Casale Monferrato in Italy, a city with an asbestos plant located as in Aalborg.5

The few previous studies reporting on non-occupational asbestos exposure in childhood are inconsistent.10 Some studies reported asbestos exposure to be more harmful when exposed in the younger ages,23 24 while Reid et al found a lower mesothelioma rate in those exposed as children compared with those first exposed at >15 years of age.25 In our study the median age at diagnosis was not significantly different in the two cohorts. Furthermore, no significant difference was found in the median age at diagnosis between those in the school cohort only environmentally asbestos-exposed and those who also had been occupationally asbestos-exposed. This might indicate that the environmental asbestos exposure is substantial enough to cause mesothelioma, and for some occupationally asbestos-exposed cases it could be a reflection of very brief occupational exposure (see table 4). Alternatively, age at diagnosis might not depend on either level of exposure or time of exposure.

As for the diagnosis of mesothelioma, the more than 70-year-old DCR was used to identify all subjects diagnosed with mesothelioma in the study period.12 The asbestos cement plant was an important workplace in the city of Aalborg, with more than 8000 employees in the period between 1928 and 1984.26 This could have caused a greater focus and awareness towards asbestos-related diseases, which may have led to bias in terms of more people being diagnosed with mesothelioma in Aalborg compared with other cities. In a subanalysis, the proportion of the school cohort mesothelioma cases who at the end of follow-up lived in the Aalborg area (0.4%) was not significantly different from those who did not live in the Aalborg area at the end of follow-up (0.2%) (p=0.08). In view of the small number of cases, these results are difficult to interpret; however, it gives an indication that bias from living in Aalborg at time of diagnosis is not present to a large extent.

In this study population, all children have been assigned to a school district according to the distance to their living address. Therefore, we used the seventh-grade school exposure as a proxy to the overall environmental asbestos exposure because it captures both the exposure in school time and the residential asbestos exposure. Residence close to an asbestos factory was also used as proxy to environmental exposure in previous studies. Newhouse and Thompson reported a relative risk of 2.2 for subjects who lived within half a mile (~805 m) of an asbestos factory in the London area.27

In a follow-up study of children living close to the crocidolite mine of Wittenoom, Reid et al found an increased incidence and mortality of MM among ‘former Wittenoom children’ compared with the West Australian population.28

A larger proportion of the school cohort had been occupationally asbestos-exposed and/or had relatives occupationally asbestos-exposed, which is not surprising given the many people living nearby, and thus having a potential for employment at the asbestos cement plant or at the shipyard in Aalborg.

Defining environmental exposure was based on exclusion of occupational asbestos exposure by means of a JEM. However, misclassification might be introduced involving the JEM. Analysing for risk of MM by using lower or higher proportions of asbestos workers to define an asbestos workplace did not change the risk estimate for neither environmental or occupational asbestos exposure. We therefore think our risk estimate for developing MM by attending school nearby the asbestos plant is valid and not skewed by exposure misclassification.

In the school cohort the male to female ratio for the environmentally asbestos-exposed cases was 1.2:1. This male to female ratio might be influenced by residual confounding. Consistent with the male to female ratio found in our study, Magnani et al reported in a study from Casale Monferrato, Italy, a male to female ratio of 1.2:1 (35 men and 29 women) for those not occupationally or paraoccupationally asbestos-exposed.29 A higher male to female ratio of 1:2.3 for a population with non-occupational exposure was reported by Marinaccio et al.4

In the comparison cohort 22 cases (male to female ratio of 2.1:1) had no apparent occupational nor familial occupational asbestos exposure. The higher number of male cases with unknown asbestos exposure could indicate that some expected misclassification from the use of the JEM has happened.

The strengths of this study are the large population size, the long follow-up time and the validated relevant information that was gathered and available through nationwide high-quality registries.

Occupational and familial occupational asbestos exposure estimation by linkage to a JEM could be considered both a strength and a limitation of the study. Our JEM estimates the historical occupational asbestos exposures avoiding recall bias. However, due to the heterogeneity of both exposure and probability of exposure within jobs, together with lack of information on employments before April 1964 and self-employments, misclassification of asbestos exposure is unavoidable, but unlikely to explain our overall results. The performed sensitivity analyses did not reveal large changes in the results, which suggests that the misclassification in our study may be limited.

MM is still a disease of importance since the incidence in Denmark is increasing despite cessation of the use of asbestos in the 1980s.30 In Denmark, patients with MM are entitled to financial compensation if they have been either occupationally or familial asbestos-exposed, but no compensation is granted if the exposure has only been environmental. In the Netherlands, it is possible to apply for compensation if either the asbestos exposure has been work-related or non-work-related, for example, environmental asbestos exposure.31 In France, the Social Security Law of 2000 created the fund of indemnification, which gives compensation to all asbestos victims—also those non-occupationally asbestos-exposed.32


In conclusion, our study adds to the evidence that environmental asbestos exposure by school attendance near an asbestos cement factory significantly increases the risk of MM. Even though occupational asbestos exposure is a powerful determinant of MM risk, our results suggest that childhood environmental asbestos exposure is also an important risk factor. In light of our results, we suggest considerations towards altering the Danish compensation regulations to apply for all mesothelioma cases regardless of type of asbestos exposure.


The authors thank the Aalborg City Archives, the Danish Civil Registration System, the Danish Cancer Registry and the Danish Supplementary Pension Fund Register for providing data used in the study. We also thank Dr Vasiliki Panou for helping with data on female mesothelioma cases.



  • Contributors ØO designed the study. JH collected, cleaned and coded raw data from registries. JH constructed the asbestos JEM. SBD performed the statistical analyses and drafted the manuscript in collaboration with ETW and ØO. All coauthors have assisted with interpretation of the findings and revised the manuscript critically. The final version has been approved by all authors.

  • Funding This study has been financially supported by Aalborg University.

  • Competing interests None declared.

  • Patient consent Not required.

  • Ethics approval The study has been performed in accordance with the Helsinki Declaration and approved by the Danish Data Protection Agency (j no: 2016-41-4787).

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