- Eero Pukkala, epidemiologist ()a,
- Rafael Aspholm, airline pilotb,
- Anssi Auvinen, professorc,
- Harald Eliasch, medical consultantd,
- Maryanne Gundestrup, senior registrare,
- Tor Haldorsen, epidemiologistf,
- Niklas Hammar, epidemiologistg,
- Jón Hrafnkelsson, medical doctori,
- Pentti Kyyrönen, statisticiana,
- Anette Linnersjö, statisticianh,
- Vilhjálmur Rafnsson, professorj,
- Hans Storm, medical doctork,
- Ulf Tveten, senior scientistl
- a Finnish Cancer Registry, Institute for Statistical and Epidemiological Cancer Research, Liisankatu 21 B, FIN-00170 Helsinki, Finland,
- bFinnish Airline Pilots' Association, FIN-01530 Vantaa, Finland,
- cSTUK, Radiation and Nuclear Safety Authority, FIN-00881 Helsinki, Finland,
- dSwedish SAS, S-19587 Stockholm, Sweden
- eNational Clinic of Aviation Medicine, Heart Centre, University Hospital, DK-2220 Copenhagen N, Denmark
- fKreftregisteret, Institute of Population-based Cancer Research, N-0310 Oslo, Norway
- gInstitute of Environmental Medicine, Karolinska Institute, S-17177 Stockholm, Sweden,
- hDepartment of Epidemiology, Stockholm Centre of Public Health, S-17176 Stockholm, Sweden
- iDepartment of Oncology, Landspitalinn, ICE-105 Reykjavik, Iceland
- jDepartment of Preventive Medicine, University of Iceland, ICE-105 Reykjavik, Iceland
- kCancer Prevention and Documentation, Danish Cancer Society, DK-2100 Copenhagen Ø, Denmark
- lInstitute for Energy Technology, N-2007 Kjeller, Norway
- Correspondence to: E Pukkala
- Accepted 23 April 2002
Objective: To assess the incidence of cancer among male airline pilots in the Nordic countries, with special reference to risk related to cosmic radiation.
Design: Retrospective cohort study, with follow up of cancer incidence through the national cancer registries.
Setting: Denmark, Finland, Iceland, Norway, and Sweden.
Participants: 10 032 male airline pilots, with an average follow up of 17 years.
Main outcome measures: Standardised incidence ratios, with expected numbers based on national cancer incidence rates; dose-response analysis using Poisson regression.
Results: 466 cases of cancer were diagnosed compared with 456 expected. The only significantly increased standardised incidence ratios were for skin cancer: melanoma 2.3 (95% confidence interval 1.7 to 3.0), non-melanoma 2.1 (1.7 to 2.8), basal cell carcinoma 2.5 (1.9 to 3.2). The relative risk of skin cancers increased with the estimated radiation dose. The relative risk of prostate cancer increased with increasing number of flight hours in long distance aircraft.
Conclusions: This study does not indicate a marked increase in cancer risk attributable to cosmic radiation, although some influence of cosmic radiation on skin cancer cannot be entirely excluded. The suggestion of an association between number of long distance flights (possibly related to circadian hormonal disturbances) and prostate cancer needs to be confirmed.
What is already known on this topic
pilots are occupationally exposed to cosmic radiation and other potentially carcinogenic elements
In the studies published so far, dose-response patterns have not been characterised
What this study adds
No marked risk of cancer attributable to cosmic radiation is observed in airline pilots
A threefold excess of skin cancers is seen among pilots with longer careers, but the influence of recreational exposure to ultraviolet light cannot be quantified
A slight increase in risk of prostate cancer with increasing number of long haul flights suggests a need for more studies on the effects of circadian hormonal disturbances
Airline pilots are occupationally exposed to ionising radiation, with doses of 2-6 mSv per year compared with an annual dose from background gamma radiation of approximately 1 mSv. Cosmic radiation at the common cruising altitudes consists mainly of gamma and neutron radiation. In 1990 the International Commission on Radiological Protection recommended that in-flight exposure of jet aircrew to natural radiation should be regarded as an occupational exposure.1 Flight personnel may also be exposed to electromagnetic fields from cockpit instruments, jet fuel, and substances emanating from materials used in aircraft construction. In addition, it has been suggested that disruptions in sleep-wake cycles associated with flying across time zones may increase the risk of cancer by suppressing secretion of melatonin2 or by some other hormone related mechanism.
The aim of this paper is to describe cancer incidence among male commercial airline pilots from all five Nordic countries. The presentation of the results and discussion about possible causes will focus on skin cancer and cancers hypothesised in advance to have an association with cosmic radiation.
We identified national cohorts of airline pilots from various registers in the Nordic countries. In Denmark we enrolled all members of commercial cockpit crew on file since 1946 in the NationalClinic of Aviation Medicine, University Hospital, Copenhagen (3790 men).3 In Finland we included all 793 male pilots who had been flying for Finnair or its subsidiaries up to 1996. In Iceland the cohort consisted of 239 men who, according to computerised files of the Icelandic Aviation Authority, were licensed commercial pilots.4 In Norway the cohort included all 3752 male commercial pilots who had a valid licence between January 1946 and February 1994.5 The Swedish cohort consisted of 1478 male aircraft pilots of the Scandinavian Airline System (SAS) resident in Sweden, who were employed between 1957 and 1994.6
We linked the cohorts to the national population registers and obtained the correct personal identifier and the possible dates of emigration, immigration, or death for every cohort member. We followed up the cohort for incident cancer cases through record linkage with the national countrywide cancer registries existing in all Nordic countries. The ethical committee of Karolinska Institute (Sweden) and the scientific committee of Copenhagen and Fredriksberg Communes (Denmark) approved the study. The national cancer registries also have an internal evaluation procedure whereby ethically unacceptable studies do not get permission to use the data.
Follow up for cancer for each participant started at the date of first employment, at immigration, or on the date of the start of cancer registration, whichever was latest, and ended at emigration, at death, or on a common closing date (the date until which cancer registration is complete), whichever was first. The cancer follow up period was 1943 to 1996 in Denmark, 1953 to 1997 in Finland, 1984 to 1997 in Iceland, 1962 to 1996 in Norway, and 1961 to 1996 in Sweden. In evaluating exposure to radiation, we also took into account the flight histories of the pilots from the years before the period of follow up for cancer.
We collected the numbers of block hours (that is, time from departure gate to arrival gate) in different types of flights from the seniority lists and other documents on flight histories. An expert panel classified the aircraft into low altitude, intermediate distance, and long distance categories, similarly in all countries. To quantify the radiation exposure, we converted the aircraft specific block hours to effective doses (mSv), on the basis of a Norwegian estimation of average radiation in different types of aircraft.7
We counted the numbers of observed cases of cancer and person years at risk by five year age groups and five year calendar periods. We calculated the expected numbers of cases for total cancer and for specific types of cancer by multiplying the number of person years in each stratum by the corresponding national cancer incidence rate.
To calculate the standardised incidence ratios, we divided the observed numbers of cases by the corresponding expected numbers. We defined the exact 95% confidence interval for each ratio on the assumption that the number of observed cases followed a Poisson distribution. We used Poisson regression to adjust for the effects of calendar period and broad age categories (<60, <=60), to assess possible interactions between the factors, and to estimate the statistical significance of the (log) linear trends.
The cohort consisted of 10 032 men under follow up, amounting to 177 244 person years (table 1). The mean length of follow up was thus 17 years. Almost 30% of the person years were in the category of at least 20 years' follow up since the time of first employment.
During the follow up, 466 cases of cancer were diagnosed compared with 456 expected, giving a standardised incidence ratio of 1.02 (table 2). The standardised incidence ratio was increased for various categories of skin cancer. The relative risk of skin melanoma increased significantly with increasing estimated exposure to radiation during flight; this trend was especially strong (P for linear trend=0.02) for skin melanoma of the trunk among pilots with >20 mSv exposure (table 3). Some indication of increasing risk with increasing radiation dose existed for other types of skin cancer and overall cancer but not for any subcategory of leukaemia or any other cancer site (table 3).
In the Poisson regression analysis the relative risk of skin melanoma increased with the cumulative dose (P for linear trend=0.008) (table 4) and decreased towards more recent calendar periods. In comparison with the reference period 1994-7,the relative risks were 1.71 (95% confidence interval 0.77 to 3.80) for the period 1943-82, 1.62 (0.72 to 3.64) for 1983-7, and 1.16 (0.54 to 2.49) for 1988-93.
The relative risk of prostate cancer in pilots aged over 60 increased with the number of block hours in long haul aircraft: eight cases occurred among pilots with more than 10 000 block hours—relative risk 3.88 (1.26 to 11.9) in comparison with the category of 1-4999 hours (P for linear trend=0.01). We observed no increase in risk with the number of block hours in long haul aircraft in younger pilots, in whom hereditary factors are assumed to have a major role.
We expected pilots to have a cancer pattern typical of high social classes 8 9 —for example, standardised incidence ratios of 1.2 for prostate cancer and 1.3 for acute myeloid leukaemia. We found ratios of 1.2 and 1.4 for these two cancers. The main exceptions upwards from the social class I pattern were various types of skin cancer. The standardised incidence ratio for skin melanoma in the highest social class varies from 1.1 in the head and neck region to 1.8 in the limbs, whereas the pilots had ratios of 2.3-2.5 across all subsites. Standardised incidence ratios of 1.5 for basal cell carcinoma and 1.1 for other non-melanoma skin cancer would be typical for social class I; we found ratios of 2.5 and 2.1. The lack of increased risk for cancers for which active case finding tends to increase incidence (such as cancers of the thyroid and prostate) suggests that regular medical surveillance is not likely to contribute to our findings. Two non-Nordic cohort studies of cancer mortality among male airline pilots 10 11 and two studies of both incidence and mortality 12 13showed excess mortality from skin melanoma (combined standardised mortality ratio 2.0, 95% confidence interval 1.0 to 3.8) and brain cancer (1.5, 0.9 to 2.2) and increased incidence of prostate cancer (standardised incidence ratio 1.7, 1.2 to 2.3) and brain cancer (1.7, 0.9 to 3.3).14
We based our estimates of dose of cosmic radiation on calculations for the Norwegian cohort,7 and we used the CARI software developed at the US Federal Aviation Authority. CARI is based on a combination of in-flight measurements and model based calculations. The results of CARI correlate with empirical measurements. Our calculations were performed for Norwegian airline pilots, but because flight distributions are comparable the dose estimates are also applicable to the other Nordic countries as semiquantitative estimates.
The effects of occupational exposure to ionising radiation on leukaemia risk have been studied primarily among cohorts of workers in the nuclear industry. A pooled analysis of seven cancer mortality studies in nuclear workers, with 2.1 million person years of follow up, showed an increased risk ratio of 1.22 for leukaemia excluding chronic lymphatic leukaemia for a cumulative protracted dose of 100 mSv compared with zero dose.15 In our cohort the estimated cumulative doses, however, exceeded 20 mSv for only 11% of the follow up time (table 1), so the excess risk expected on the basis of the previous risk estimates would not exceed 1.05. We found only seven cases of non-chronic lymphatic leukaemia in pilots with a cumulative dose of 3 mSv (4.2 expected), and no clear association with radiation dose existed.
Consistent with our finding of an increased risk of skin cancers, the three Nordic studies on incidence of cancer among cabin crew published so far indicate a twofold to threefold excess risk of skin cancer.16–18The major risk factors for malignant melanoma of the skin include intermittent sun exposure, sunburn at early age, and host factors related to skin colour and naevi.19Aircraft windows block ultraviolet radiation, and measurements of ultraviolet radiation reveal only minimal exposure in the cockpit at high cruise altitudes.20
We had two a priori approaches to separate the behavioural risk factors for skin melanoma from direct occupational hazards such as cosmic radiation. Firstly, because the relative difference between pilots and other people in the frequency of travelling to the south is diminishing over time, we hypothesised that the standardised incidence ratio would decrease towards recent years if it is related to sunbathing. The relative risk for skin melanoma among pilots was clearly higher before the mid-1980s than in the 1990s, suggesting that sunbathing is a major determinant of risk. The second approach was to study the risk by skin area: the risk in the head and neck area is similar irrespective whether the person is regularly outdoors or not, whereas skin melanoma in the trunk and limbs is much more common among indoor workers.8 The risk among pilots was similarly increased for melanoma in the head and neck and for melanoma in the trunk and limbs, which according to our hypothesis might point to factors other than sunbathing.
In previous studies, skin melanoma has rarely been associated with ionising radiation. We conclude that although the relative risk of skin melanoma increased significantly with estimated dose of ionising radiation, the excess may well be attributable to ultraviolet radiation.
The causes of prostate cancer are poorly understood. Melatonin may have a protective effect,2 and disruptions to sleep-wake cycles may increase cancer risk by suppressing melatonin secretion. According to one hypothesis, melatonin would increase the risk of hormone related cancers, whereas other people believe that it would have a systemic antioxidant-like effect. According to this theory, flight personnel flying over many time zones might have an excess risk. The main hormone related cancer among men is prostate cancer. Among pilots in our study, the relative risk of prostate cancer increased with the number of long haul hours. We cannot fully exclude the possibility that jet lag would increase the risk of hormone related cancers.
Few areas outside the Nordic countries have a history of several decades of population based registration of cancer. The large joint cohort size and five independent country specific observations in this joint Nordic study reduce the possibility of chance findings, and carefully registered incidence data help to avoid artefacts possibly included in the mortality statistics. Because of the complete population registration systems in all Nordic countries and accurate computerised record linkage procedures, the standardised incidence ratio estimates in this study are not prone to bias attributable to incomplete follow up or failures in record linkages.
Because our cohort included most of the cockpit crew ever certified in the Nordic countries, this study can be considered to have maximum potential to study cancer incidence among pilots. We were able to assess cancer risk by level of exposure, take into account characteristics of cancer (for example, subtypes of leukaemia, tumour latency), and estimate independent effects of exposure, age, and time period of the diagnosis in a way that has not been meaningful in small national settings. Apart from skin cancers, male Nordic pilots seem to have a pattern of cancer typical of that of high social class men in the Nordic countries. Our study shows a need for detailed studies focusing on possible work related factors involved in the evidently raised risk of skin cancer and the suggestive dose-response patterns in prostate cancer.
Contributors: All the authors actively participated in design, planning of analysis, writing the paper, and agreeing about conclusions. EP coordinated the work and is the guarantor.
Funding Nordic Cancer Union and the participating institutes
Competing interests RA has been employed by Finnair as an airline pilot since 1988 and has shares in Finnair. He is also an active member of the Finnish Pilots' Association and has been reimbursed by the association for attending medical symposiums and conducting scientific research. HE has worked as an medical consultant for Scandinavian Airlines. The other authors have no connections to airline companies.