Discussion

This is the first large study of cancer incidence among firefighters in the Nordic countries. The overall cancer incidence was close to that of the general population, but statistically significant excesses were observed for several cancers.

The pattern of excess cancer incidence observed in this study is at least partly consistent with patterns observed in previous studies (table 5). Although small, the excesses of malignant melanoma, non-melanoma skin cancer and prostate cancer have been observed in many other studies.1 Additionally, the small, non-significant excesses of stomach cancer, colon cancer and multiple myeloma seen in our study were also identified in the meta-analyses of LeMasters et al.5 By contrast with the IARC1 and LeMasters et al5 comprehensive reviews, our study observed significantly fewer testicular cancers than predicted. A similar finding of reduced testicular cancer incidence (SIR 0.74, 95% CI 0.41 to 1.22) from the simultaneous study of almost 30 000 US career firefighters employed since 1950 and followed through 2009,12 makes it less likely that our observation would be due to chance only. The aetiology of testicular cancer is unknown, and hence, we cannot give any explanation on this finding.

This study provides only little support for the meta-analysis result for non-Hodgkin lymphoma,5 since the SIR for this cancer was above 1.0 only among firefighters over the age of 70 years. Care should be taken when comparing observations derived from different studies, because the calendar periods of exposure and follow-up, age distribution of the years at follow-up and even the levels of reference rates used in comparison greatly affect the relative risk estimates.

Although the overall excesses were modest, there were quite high relative risks either in the youngest or oldest age categories of follow-up. The more than twofold excess of prostate cancer in ages below 50 years observed in this study is of particular interest because of recent concerns regarding shift work involving the disruption of circadian rhythms.1 Although breast cancer has been the main focus of study for shift work, all four studies of acceptable quality on prostate cancer among night workers show an almost twofold increased risk.13–16 Additionally, studies of flight crews have also identified greater than expected numbers of prostate cancers.1 ,17 It is not possible, based on existing studies, to qualify whether some type of shift is worse than others in terms of adding a risk of prostate cancer. We do not have statistics of shift types of firefighters in various Nordic countries in the past, but it is likely that they vary between countries and even within countries. In Denmark and Finland, it has been common to have 12 h and 24 h shifts. It is, however, important to keep in mind that the majority of firefighters normally sleep during the night, except if fires occur. We do not know about the quality of firefighters’ sleep, but they probably sleep less than ordinary day workers (which in itself is suggested as contributing to prostate cancer) and have lower quality of sleep.

One review has also suggested there may be evidence for an association between prostate cancer and polycyclic aromatic hydrocarbons and diesel exhaust.18 The excess observed among relatively young men is intriguing and should be examined in other cohorts. Daniels et al12 report in their paper that excess of prostate cancer was limited to ages of 45–59 years (SIR=1.44, 1.27 to 1.63), but they kindly calculated the SIR for the same age category that was used in our study, that is, 30–49 years. Their SIR rate for prostate cancer in this age range is 2.04 (95% CI 1.43 to 2.82), which is similar to the rate seen in our cohort (SIR 2.59, 95% CI 1.34 to 4.52). This finding is new. One explanation of the excess incidence might be that the firefighters would be targeted for prostate-specific antigen (PSA) testing in ages below 50 years more often than other men in the same ages.

No documentation is available on this point, but occupational health experts we have consulted do not expect this to be the case. The SIR of prostate cancer in ages 30–49 years did not change after the frequency of PSA testing started to rapidly increase in the 1990s which further increases the possibility that the finding might not be due to surveillance bias.

This study observed a small excess of malignant melanoma of the skin and non-melanoma skin cancer. These results are consistent with a number of previous studies of firefighters.5 Skin cancer has been associated with a number of polycyclic aromatic hydrocarbon-related exposures, such as soot, coal tar pitch volatiles and unrefined mineral oils.19 The combustion of any organic materials, as may be encountered by firefighters, are likely to contain polycyclic aromatic hydrocarbons. Associations between occupational exposures and malignant melanoma are not well established. In our stratified analyses, malignant melanoma and non-melanoma skin cancer show opposite relationships with age (skin melanoma elevated in younger firefighters, non-melanoma skin cancer in older groups) and time period (skin melanoma elevated in the earliest time period, non-melanoma skin cancer in a later time period).

The large US firefighter cohort study12 could only give rates for skin melanoma. In their cohort, the SIR was low in the age category of 30–59 years (SIR 0.44, 95% CI 0.22 to 0.79) and close to 1.0 in the older age categories. We, in turn, observed an excess of skin melanoma in the age range of 30–59 years (SIR 1.62, 95% CI 1.14 to 2.23) and SIRs close to 1.0 in the older age categories. Registration of skin cancers in the USA may be less complete than for other types of cancer. Furthermore, in the case of skin melanoma (with large geographic variations in incidence), the problem of having the firefighter cohort mainly from California but reference incidence rates from the entire Surveillance, Epidemiology, and End Results (SEER) population might cause bias, and the US results should be interpreted with caution. In the Nordic countries, an elevated risk of skin melanoma has been observed among persons of quite high socioeconomic position, such as airline pilots, for whom tanned skin has been a symbol of a healthy appearance.17 The sun exposure of the firefighters is not known, but it could potentially play a role.

This is the first study to identify an excess of mesothelioma among firefighters. Consistent with its long latency, a significant excess was observed in older firefighters. This may indicate a continuing hazard faced by firefighters encountering asbestos remaining in building materials. Asbestos exposure has long been identified as a potential concern among firefighters, but previous studies may not have had sufficient weight or enough long follow-up to identify a small, but important excess of this cancer. One of the reasons for the lack of previous reports on mesothelioma is simply that nearly all previous studies examined mortality using the underlying cause of death as coded according to ICD versions prior to ICD-10. A specific mesothelioma code was not available until the 1990s with the publication of ICD-10. The SIR for mesothelioma was highest in Norway (table 2), but the differences between the country-specific rates are not statistically significant. We do not have any information suggesting that asbestos exposure in Norway would be higher than in the other countries. The SIRs in the US data by Daniels et al12 for the age category of 70+ was 2.44 (95% CI 1.55 to 3.67), which is almost identical with the respective rate in our Nordic cohort (SIR=2.59, 95% CI 1.24 to 4.77). Daniels also observed an excess risk of mesothelioma in the age category of 50–69 years (SIR 2.22, 95% CI 1.14 to 3.87) while we did not. Such a difference may well be related to different patterns and time periods of asbestos use in the buildings in California and the Nordic countries.

Although the lung would seem a likely target organ for firefighters exposed to combustion products and diesel exhaust, excesses of respiratory cancer have only rarely been observed. One exception is the earlier Danish study,6 which observed an excess of lung cancer among older firefighters. Our study also observed an excess of lung cancer, which was only among Danish firefighters and mainly due to the high relative risk of adenocarcinoma of the lung among older firefighters. The other two large cohort studies of firefighters to examine incident cancer did not report analyses by histologic type.12 ,20

Our study was based on a large cohort with a long follow-up period. Due to the high coverage, precision and validity of the linked files, the cancer risk estimates can be considered reliable. Because the present study was based on incident cancer cases and exact person-years, there was no bias caused by occupational variation in cancer survival and in mortality from competing causes of death.

The occupational affiliation at one point in time may not always correspond to the lifelong occupational history of a person. Comparison of results based on only one cross-sectional information on occupation with results from studies with complete occupational histories indicates that the diluting effect due to misclassification is small especially in specialised occupations.8 The average age for firefighters to finish their education in Finland in 2005–2013 was 24.7 years. The occupational stability is high, and up to the age of 55 years, more than 80% of the men pass the physical test and stay in the operational firefighter work.

This study on firefighters, however, has several specific limitations that should be considered. Information on duration of employment and other characteristics of exposure was not available. Assessing firefighters’ exposure is a difficult challenge for any study. Only two studies have gone beyond simple duration and created metrics based on the numbers of fires fought.7 ,21 However, even the number of fires may not always be a good surrogate for actual exposure.22 As with all other retrospective studies of firefighters, our ability to control for confounding was very limited. Thus, characteristics of firefighters, such as lifestyle factors that differ with the general population, could confound our results. For example, if the study population had higher levels of non-occupational sun exposure, it could increase their risk of skin cancer.

The firefighters probably represent a selected population group in terms of general health conditions. Firefighters are mainly employed by the municipalities and may be required to undergo a health check before recruitment, and an extensive health check is required for participation in firefighter team leader courses. We know some of the characteristics of Finnish firefighters from surveys on the health habits of the Finnish adult population during 1978–1991 collected for the Finnish Job Exposure Matrix.23 The prevalence of smoking among firefighters was 29%, which is lower than in the general male population (33%). Weekly alcohol consumption was 112 g (population: 84 g), and body mass index (BMI) 26.1 (25.4). These data suggest that firefighters follow a quite similar way of life than the reference population, and therefore our cancer risk estimates should not be strongly confounded by lifestyle factors. In a recent study on cancer risk of male world class athletes,24 the BMI of the athletes was not considered to indicate overweight in the same way as among average persons and was not included in the analyses, because the BMI for athletes is elevated due to muscular build instead of body fat.25

This study has several important strengths compared to previous studies. In terms of numbers of cases, this is larger than any cohort study of firefighters published before 2014. Thus, this study had the capability to identify small excesses, even for relatively rare cancers. Additionally, this study had access to high-quality cancer registry data from across the Nordic countries. The cancer incidence data allowed for better identification of non-fatal cancers and those identified on the basis of histology (such as mesothelioma), as well as allowing the examination of histologic subtypes of cancer.26

In summary, this study suggests that firefighters may be at increased risk of prostate cancer, malignant melanoma, non-melanoma skin cancer, multiple myeloma, mesothelioma and lung adenocarcinoma. Some of these associations have been observed previously, and potential exposure to polycyclic aromatic hydrocarbons, asbestos and shift work involving disruption of circadian rhythms may explain some of these associations. However, further studies of firefighters with better indices of exposure and, ideally, more individual-level information on other risk factors for cancer are needed to more firmly explore the aetiology of these cancers among firefighters.