Two cohort studies on cancer risk in firefighters were recently published in OEM.1 ,2 One was a cohort of firefighters based on employment records from three US cities which had been linked with State Cancer Registries and the National Death Registry, and the other was a cohort drawn from census data from five Nordic countries which had been linked with each country's National Cancer Registries. They were both well-conducted cancer incidence studies which included large numbers of firefighters and had long follow-up periods. There were small but statistically significant increases in standardised incidence ratios (SIRs) for all cancers in both studies (SIR=1.10 95% CI 1.07 to 1.13 in the US study and 1.06 95% CI 1.02 to 1.11 in the Nordic study).

A new finding that emerged from both studies was an increase in mesothelioma. In the US study, the SIR was 2.29 (95% CI 1.60 to 3.19) and in the Nordic study it was 1.55 (95% CI 0.90 to 2.48). This increase in risk is likely to be because of the asbestos exposure occurring when buildings burn, during clean up and also perhaps a result of the asbestos protective gear which used to be widely worn by firefighters.3 These are the first studies to show a statistically significant increased risk of mesothelioma for firefighters. Worldwide asbestos production and use rose rapidly from 573 728 metric tons in 1940 to 3 493 800 in 1970.4 Firefighter exposure will therefore have become more likely over this period, and because of the long latent period between asbestos exposure and mesothelioma development, which may take over 40 years,5 the probability of contracting mesothelioma would be more likely to be identifiable in recent cohorts. Although asbestos is no longer used in new buildings in either of the locations studied, the persistence of asbestos in old buildings will continue to expose firefighters to this risk.

Both of these cohorts showed significantly decreased risks of testicular cancer (SIR=0.51 95% CI 0.23 to 0.98 for the Nordic study with 9 cases and SIR=0.62 95% CI 0.36 to 0.99 for the US study with 15 cases). A previous meta-analysis based on two incidence studies found an increase in testicular cancer risk (summary risk estimate (SRE)=1.83 95% CI 1.13 to 2.79)6 and a later review of six studies (including a mortality study) found an SRE of 1.47 (95% CI 1.20 to 1.80).7 Testicular cancer is a rare cancer and numbers in the cohort studies as well as the meta-analyses were relatively small. Testicular cancer has good survival, so including the mortality study in the meta-analysis may not be appropriate as rates measured in mortality studies may not be comparable with rates from incidence studies. In addition, few studies report data on this rare cancer, so there may be some publication bias operating. Therefore, the significantly decreased risks in the two recent cohorts may just be chance events.

Similarly, increased risks of non-Hodgkin lymphoma (NHL) which had been noted in the meta-analyses6 ,7 were not seen in the two cohorts. For NHL, the US study showed an standardised mortality ratio of 1.17 (0.97 to 1.40) and a lower SIR of 0.99 (0.84 to 1.15) while the Nordic SIR was 1.04 (0.83 to 1.29). In the studies included in the meta-analyses, standardised effect measures for NHL in earlier studies tend to be higher than those in recent studies. All three studies published after 2000 and included in the meta-analyses had non-statistically significant findings for NHL.8–10

On the other hand prostate cancer, which the meta-analyses found was increased in firefighters,6 ,7 was also increased in the two new cohorts. Subanalyses showed that the increase tended to be in younger men and in the US cohort was restricted to only the San Francisco department and to non-European Americans, while in the Nordic study the rate increased over time with the highest rate in the most recent time period. Prostate cancer is quite a common cancer and so the consistent findings from the meta-analyses as well as the two new cohorts strengthen the conviction that this finding is causal.

Melanoma, which had not been increased in previous meta-analyses,6 ,7 was seen to be increased in the two firefighter cohorts. In the Nordic study the increase was seen in younger cohorts, but has decreased over time, while in the US study the increase was confined to the San Francisco department. One of the theories about the association between sun exposure and melanoma is that sunburns, particularly but not only, before the age of 15, increase the risk of melanoma more than cumulative exposure.11 These two cohorts may provide some support for this theory in that perhaps the opportunity for sunburn was higher in younger firefighters with the increasing travel to sunny holiday destinations, and perhaps greater in southern San Francisco than in more northern regions of the USA. There are no other known occupational exposures in firefighting which are linked to melanoma.

The inconsistencies between the meta-analyses and these two very large cohort studies are interesting and may be because firefighting is a diverse occupation with a range of different exposures which have changed over time. Neither of the two new cohorts, nor any of the previous studies have been able to assess specific exposures for individuals in the cohort. Firefighter exposures in the different studies may differ because of a number of factors. First, the types of fires which are fought will give rise to different exposures. The predominant fuel in fires in rural locations is wood and organic matter. In contrast, firefighters in urban locations may be exposed to a wide range of potential carcinogens arising from many different flammable materials in homes, office buildings, industrial sites, vehicles and waste dumps. In addition, urban firefighters may be involved in clean ups after industrial or traffic accidents and chemical spills, as well as explosions.12 ,13 Second, there have been changes over time in the materials which may burn, including a recent increase in the use of combustible plastics.14 Third, there have been changes in the type and use of respiratory protection. Research has shown that backburning and overhaul may give rise to as much or more exposure than knockdown,15 ,16 and firefighters may not have used respiratory protective equipment for backburning and overhaul in the past.17 ,18

The two cohort studies included fire fighters employed from the 1950s (USA) and 1960s (Nordic) as well as some employed recently. Because of the changes in exposures, work practices type and use of protective clothing and respiratory protection noted above, the recent hires would be exposed to different levels and types of carcinogens.

From a risk reduction point of view, firefighters should be encouraged to reduce their risk of exposure to smoke and other pollutants by avoiding exposure where possible. Protective clothing and respiratory protection should be used during training, knockdown, overhaul and backburning. Studies have shown that breathing apparatuses have not been consistently used in the past even for firefighters attending structural fires.15 ,18 ,19 There is some evidence that the skin can be a route of exposure and so skin contact should also be avoided.20 The newer protective clothing has contributed to the reduction in burn injuries so it is likely that it would also reduce skin contact with pollutants.21

These studies have provided some further information about the extent to which firefighters are at greater risk of cancer; in particular, they have strengthened the evidence regarding increased risks of prostate cancer and mesothelioma. However, prevention in future will depend on having better exposure data integrated into the epidemiology in order to identify the exposures that are associated with increased risks. The lack of such data is a persisting limitation which has been previously highlighted in the International Agency for Research in Cancer (IARC) Working Group and the meta-analysis.6 ,7 This has a parallel with other carcinogenic occupations considered by IARC, such as welding and painting where there is no identification of any specific exposure which is associated with the risk of cancer.