Risk of cancer associated with residential exposure to asbestos insulation: updated evidence

All forms of asbestos are classed as known human carcinogens.¹ More than 80% of pleural mesothelioma cases in men are due to asbestos exposure.² Although the threat of asbestos is well-established in occupational settings, less is known about the risk to residents of homes fitted with asbestos insulation, in particular of the loose fill variety.

Between 1968 and 1979, a private company installed loose fill asbestos insulation into the roofs of over 1000 homes in the Australian Capital Territory (ACT) region. Amosite asbestos was used in most homes, although crocidolite was later found in two homes.³ Despite initial efforts to remove visible and accessible asbestos, fibres were found in the living spaces of many affected homes more than 20 years later. In some cases, asbestos fibres were visible on household furnishings. By August 2014, the ACT government decided that all affected homes were to be demolished.³

In 2015, our research team conducted a cohort study to examine the risks of mesothelioma and six other cancers among residents of homes with loose fill asbestos insulation in the ACT.⁴

The main finding of our initial study was that men who had lived in these homes were more likely than the rest of the ACT male population to develop mesothelioma.⁴ The study also found a higher rate of colorectal cancer among women who had lived in these homes but the same finding was uncertain in men. We have not found any other scientific reports on the risk of cancer in residents of homes with loose fill asbestos insulation since this time. In this article, we report on an update of the initial study incorporating six additional years of follow-up and increased statistical power. We also used improved data linkage methods and included additional cancer outcomes potentially caused by asbestos exposure.

The study population included all individuals in the Australian Medicare Consumer Directory (MCD) (previously Medicare Enrolment File) who had an address in the ACT at any time between 1983 and 2019, extending the previous study by six years. The MCD data, which includes nearly all of the Australian population, were linked to three databases: 1) a historical register with addresses of 1095 known affected residential properties (ARPs) in the ACT,⁵ including six additional homes that had not yet been identified in the initial study; 2) the Australian Cancer Database (1982–2019); and 3) the National Death Index (1982–2019). Data linkage between the MCD and list of ARPs was facilitated by geocoded addresses.

We calculated the incidence rates of mesothelioma (ICD-10 C45), pharyngeal (C09–C14), stomach (C16), colorectal (C18–C20), laryngeal (C32), lung (C33–C34), and ovarian (C56) cancers. Additionally, we included liver (C22) and oesophageal (C15) cancers in this study. We also examined four control outcomes: melanoma (C43) and prostate (C61), kidney (C64), and bladder (C67) cancers. We calculated the standardised incidence ratio (SIR) for each cancer outcome with exact Poisson 95% confidence intervals (CIs), separately for males and females, allowing for a 10-year lag between exposure and outcome. SIRs were derived from indirect standardisation, applying age- and calendar-period specific incidence rates from all ACT individuals who had never lived in an ARP to the exposed group (see our initial study for more detail on study methods and datasets⁴). We performed sensitivity analyses as in our initial study.⁴

Our study included a total of 16,757 individuals (1.4%) who had lived in an ARP between 1983 and 2019, and 1,207,626 individuals (99%) who had never lived in an ARP (see Appendix Figure S1 for sample selection and Appendix Table S1 for demographic characteristics). These individuals were followed for a total of 0.26 million exposed person years and 29 million unexposed person years.

There were 12 incident cases of mesothelioma among men who had lived in an ARP. The median time to diagnosis after the first recorded exposure was 26 years (Q1, Q3: 16, 30) (see Appendix Table S2 for mean and median for all cancers examined). This is a minimum estimate, as 6 of the 12 men were enrolled with an ARP address at the inception of the MCD.

We observed elevated rates for mesothelioma in men (SIR = 2.72, 95% CI 1.41–4.76), colorectal cancer in both sexes (SIR = 1.24, 95% CI 0.99–1.54 males; SIR = 1.46, 95% CI 1.15–1.83 females) and lung cancer in women (SIR = 1.39, 95% CI 1.00–1.88) (Fig. 1, see Appendix Table S3 for crude rates). We did not observe statistically significant higher-than-expected rates for all other candidate cancers examined, however there was low statistical power to detect modest associations. Sensitivity analyses did not alter our conclusions (Appendix Figures S2–S6).

Fig. 1.

Observed (O) and expected (E) case numbers in the exposed population and standardised incidence ratios (SIRs). Forest plots show point estimates of SIRs (filled squares) and 95% confidence intervals (horizontal lines). SIRs are plotted on a log scale. SIRs where there were fewer than six observed cases in the exposed group are not reported to minimise statistical disclosure risk. Note that in such instances, there was low statistical power to detect effect sizes within reasonably expected magnitudes.

We observed that the rate of mesothelioma in men who ever lived in an ARP during 1983–2019 was 2.7 times that of the rest of the male ACT population, while rates of colorectal cancer in exposed men and women, and lung cancer in exposed women, were between 1.2 and 1.5 times that of the unexposed populations. These estimates were similar, but more precise, than the initial study, in line with the longer follow-up time in the current study.

Given the strong causal link between occupational exposure to asbestos and mesothelioma, the large SIR for mesothelioma in our study supports an increasing concern from asbestos exposures in non-occupational settings.⁶ While non-occupational sources—generally neighbourhood, domestic/para-occupational and household—are distinct in their exposure routes and impact, there is emerging evidence of elevated risks of mesothelioma from all three sources.⁶ Household exposures remain the least studied, especially as an exposure type independent from domestic/para-occupational definitions. Our study provides a clear example of household exposure in relation to the risk of mesothelioma.

The additional years of follow-up time allowed a more precise estimate for mesothelioma to be calculated, and clearer communication of the risks to the affected community. The use of linked administrative data rather than primary data collection facilitated this study update, given that mesothelioma's low incidence and long latency period require a large and long-term cohort to achieve satisfactory statistical power.

We remain uncertain of the link between loose fill asbestos insulation and colorectal cancer, since the relatively small effect sizes could be potentially explained by confounding, and given a lack of convincing prior evidence. While inhaled asbestos can also be swallowed into the gastrointestinal tract,⁷ a recent umbrella review regarded that the links between occupational asbestos exposure and gastrointestinal cancers, including colorectal cancer, remain weak.⁸ Studies of colorectal cancer associated with non-occupational exposure are scant.

As for confounding, we did not have data on other risk factors (e.g., occupational exposure to asbestos or other lung irritants, smoking and socioeconomic status) which may have biased study estimates, although we had no suppositions that these risk factors were distributed differently in residents of ARPs compared to non-ARP residents. Affected residential properties were spread throughout the ACT,⁵ excluding suburbs developed after 1979. Nevertheless, we did consider adjusting for area-level socioeconomic status, but published area-based indices are a poor measure of individual-level disadvantage in the ACT.⁹ Additionally, some point estimates for control outcomes were above one, reflecting a possible higher baseline risk of cancer in residents of ARPs.

Regarding the excess rate of lung cancer observed in women who had lived in ARPs, there is strong evidence that occupational exposure to asbestos increases the risks of developing and dying from lung cancer.⁸ There is developing evidence that non-occupational asbestos exposure can also increase the risk of lung cancer, but less so for household exposure specifically.^10,11 As with other studies on asbestos-related lung cancer, the potential role of smoking, which is both a confounder and an effect modifier, is challenging to address.^12,13 We could not confidently rule out that the SIRs for lung cancer in our study may be biased without data on tobacco use and other potential risk factors.

We examined two additional cancers in this study that we did not include in the initial study—oesophageal and liver. We did not observe higher-than-expected rates of oesophageal cancer in men. There was an inadequate number of cases to report results for oesophageal cancer in women, and for liver cancer in either men or women, due to statistical disclosure requirements from the data custodian. However, in these cases, there was insufficient statistical power to detect modest associations.

Demolition of affected homes began in 2015, prior to our initial study. We were unable to estimate historical fibre concentrations, which were thought to vary widely between houses according to their age, quality of construction, the effectiveness of cleaning undertaken during an earlier clean-up program (1989–93), and other factors including ground stability.³

Asbestos remains a legacy material in homes, as well as in public and commercial buildings across Australia and the broader Asia Pacific region. Although Australia, Japan, South Korea and Taiwan have instituted asbestos bans, the active construction and demolition of buildings containing asbestos in many other countries in Asia continue to contribute to non-occupational exposure.¹⁴ While the use of friable asbestos as insulation has been less common in Asia, such circumstances leading to indoor exposure has been documented in South Korea,¹⁵ underscoring the relevance of long-term surveillance from all sources of non-occupational exposure and their particular natures.

Our updated cohort study adds to the scientific literature on residential exposure to asbestos in relation to a wide range of cancers across all age groups. Although exposure to asbestos outside the workplace are generally at lower levels, they may be chronic and include children and women rather than predominantly male workers. Our study confirms that living in a house with loose fill asbestos insulation may be sufficient to cause cancer, in particular mesothelioma.

Contributors

HDL and NL had access to the data in the study and did the statistical analysis. HDL wrote the first draft of the manuscript with input from RJK. All authors contributed to the study design, provided critical feedback and approved the final manuscript.

Data sharing statement

The data used in this study will not be made available to individuals other than nominated members of the study team under the Australian Government Department of Health requirements for data linked to the Medicare Consumer Directory.

Ethics approval

We obtained ethics approval for the study from the following institutions: ACT Health Human Research Ethics Committee (HREC) (2022.ETH.00197), AIHW Ethics Committee (EO2022-5-1387), ANU HREC (2022/623), NT Health and Menzies School of Health Research HREC (2023–4513).

Declaration of interests

The authors declare that they have no conflict of interest.