Mortality from cancer and other causes among male airline cockpit crew in Europe

Authors


Abstract

Airline pilots and flight engineers are exposed to ionizing radiation of cosmic origin and other occupational and life-style factors that may influence their health status and mortality. In a cohort study in 9 European countries we studied the mortality of this occupational group. Cockpit crew cohorts were identified and followed-up in Denmark, Finland, Germany, Great Britain, Greece, Iceland, Italy, Norway and Sweden, including a total of 28,000 persons. Observed and expected deaths for the period 1960–97 were compared based on national mortality rates. The influence of period and duration of employment was analyzed in stratified and Poisson regression analyses. The study comprised 547,564 person-years at risk, and 2,244 deaths were recorded in male cockpit crew (standardized mortality ratio [SMR] = 0.64, 95% confidence interval [CI] = 0.61–0.67). Overall cancer mortality was decreased (SMR = 0.68; 95% CI = 0.63–0.74). We found an increased mortality from malignant melanoma (SMR = 1.78, 95% CI = 1.15–2.67) and a reduced mortality from lung cancer (SMR = 0.53, 95% CI = 0.44–0.62). No consistent association between employment period or duration and cancer mortality was observed. A low cardiovascular mortality and an increased mortality caused by aviation accidents were noted. Our study shows that cockpit crew have a low overall mortality. The results are consistent with previous reports of an increased risk of malignant melanoma in airline pilots. Occupational risk factors apart from aircraft accidents seem to be of limited influence with regard to the mortality of cockpit crew in Europe. © 2003 Wiley-Liss, Inc.

Airline pilots and flight engineers are occupationally exposed to ionizing radiation of cosmic origin and to electromagnetic fields from cockpit instruments.1 Contact with jet fuel and inhalation of fumes2, 3 could also pose occupational hazards. Whether chronobiological disturbances caused by frequent crossing of time zones have any long-term health effects is currently unclear.

Among other things, the potential hazards of ionizing radiation have motivated research focussing on exposure quantification and health consequences, in particular as no human data concerning health effects of neutrons are available. Cosmic radiation originates from high-energy particles with sources in and outside the solar system. Interactions with the atmosphere lead to the creation of neutrons that contribute up to 60% to the effective radiation dose.4 Highest radiation doses occur in jet flights in high altitude and on polar routes. In-flight measurement approaches5, 6 and dose calculations7 yield rather similar results indicating that typical cockpit crew are exposed to annual radiation doses in the range 2–6 mSv.

Epidemiologic cohort studies of cancer incidence and mortality among civil aviation cockpit crew have been conducted in North America,8, 9, 10 Japan11 and European countries.12, 13, 14, 15, 16, 17, 18, 19, 20 Increased risks for malignant melanoma of the skin have been the most consistent finding to date. Risk increases for cancers of the brain, large bowel, and prostate as well as for leukemia were found in some studies, but not in all. Typical findings have included a high risk of aviation accidents and decreased mortality from cardiovascular causes.

Most studies reported so far have been relatively small. Opportunities to evaluate dose-response analyses with e.g., duration of employment or other exposure indices have been limited. To address these issues, a collaborative European mortality study among flight personnel was set up. The primary aim was to assess the mortality of flight crew in relation to occupational factors, with a focus on potential effects of ionizing radiation. This is a report on the mortality of cockpit crew, i.e., pilots and flight engineers, from 9 European countries.

MATERIAL AND METHODS

The European Study of Cancer Risks among Airline Personnel (ESCAPE) is a collaborative European investigation in which cockpit crew cohorts were collected in nine countries (Denmark, Finland, Germany, Great Britain, Greece, Iceland, Italy, Norway, Sweden) according to a joint protocol and subsequently analyzed together. The national cohorts of cockpit crew were compiled from varying sources in the participating countries (Table I). Procedures have been described previously in more detail for most cohorts.12, 13, 14, 15, 16, 18, 19 In short, cockpit crew were identified from airline personnel files in Finland, Germany, Great Britain, Greece, Iceland, Italy and Sweden, supplemented by data from airline pilot associations in Finland and Iceland. Pension records were used in Great Britain. In Norway, license data of the national aviation authority were used for cohort identification and enumeration; this source was also used in Iceland. In Denmark, all commercial cockpit crew on file in the National Clinic of Aviation Medicine in Copenhagen were included.

Table I. Description of Escape Male Cockpit Crew Cohort
CountryData sources for cohort enumerationCohort inclusion periodFollow-up periodCockpit crew (males only)Deaths
NumberPerson years
  • 1

    Decimals rounded

DenmarkNational Clinic for Aviation Medicine, Aviation Authority1946–19961961–19963,81463,090268
FinlandFinish Airline Pilots Association, Finair1921–19971971–199678212,33350
GermanyDeutsche Lufthansa, LTU1953–19971960–19976,061105,037255
GreeceOlympic Airways1946–19971960–199784317,58765
IcelandNational Aviation Authority, Airline Pilots Association, Icelandair1935–19971961–19974388,26547
ItalyAlitalia1965–19951965–19963,02054,008212
NorwayCivil Aviation Administration1946–19941962–19963,66465,317318
SwedenSAS Sweden1957–19941961–19961,40528,36291
United KingdomBritish Airways1950–19971950–19977,770193,564945
Total   27,797547,56412,251

The data collected for each individual included name, gender, date of birth and employment or license data. Wherever possible, information on detailed occupational history was recorded. For each cohort member, start of employment and, if retired, end of service as crew member were available. In several cohorts further employment data such as aircraft type and flight hours were collected and radiation exposure was estimated.21, 22, 23 These data were not used in the current analysis of the complete cohort as they are still incomplete.

Follow-up for the vital status of each individual started at the first date of employment or licensing, at immigration, or the country-specific start of follow-up, whichever was latest, and ended at date of death, date of loss to follow-up, emigration or at the end of the study period (end of 1996 or 1997, depending on country) whichever came first.

Follow-up was carried out through centralized or local population registries in most countries. In Denmark, Finland, Great Britain, Iceland, Italy, Norway and Sweden causes of deaths were established through record linkage with national death registries. In the Nordic countries, the unique personal identifier of all cohort members was recorded and the linkage done electronically. In Greece social security records were used to establish vital status; these records also contained the cause of death for deceased cohort members. Follow-up for vital status in Germany was done through population registries and local health offices provided copies of death certificates. Information from physicians or relatives was retrieved when the death certificate was not accessible. All deaths were coded according to the relevant revisions of the International Classification of Diseases (ICD revisions 7-10). Data were analyzed centrally at the University of Bielefeld, Germany. Cohort members for whom date of birth or date of beginning of employment/license were missing were excluded from the analysis. Data for female cockpit crew were not analyzed as the cohort and the number of deaths were small.

Statistical analysis

Expected numbers of deaths were computed using cause-specific population mortality rates for each country from the WHO mortality data base or supplied by the national statistical offices. The period 1960–97 was used as time window for the analysis. Person-years at risk and mortality rates were calculated in 5-year age and calendar intervals, and standardized mortality ratios (SMR) were estimated.24 Because the cause of death was unknown for some persons, a correction method proposed recently was applied that allows an improved estimation of the SMR in the presence of missing causes of deaths.25 For this correction, the observed number of cases is divided by a correction factor equation image that represents the proportion of persons with known cause of death among all deaths (0 ≤ equation image≤ 1), thus increasing slightly the observed numbers. Appropriately corrected26 95% confidence intervals (CI) for the SMR were calculated based on an approximation or on exact Poisson distribution. Heterogeneity was investigated by χ2 tests24 and by plotting country-specific SMR results. In addition an analysis of mortality stratified by time since first employment for mortality from main causes and radiation-associated cancers was conducted. In a Poisson regression analysis the effect of duration of employment/license adjusted for age (using 5-year age groups) and calendar period (categories 1960–69, 1970–79, 1980–89, 1990–97) was assessed. Further adjustment for country did not change risk estimates, but led to model convergence problems in several instances.

RESULTS

An overall cohort of 28,066 male cockpit crew and 262 female cockpit crew members was compiled. In total, 279 cohort members had to be excluded from the analysis since date of birth, sex or date of first employment were unknown. The 27,797 male cockpit crew members finally included in the study yielded 547,564 person-years. Of the total cohort, 5.3% emigrated or were otherwise lost to follow-up.

Among male cockpit crew, a total of 2,244 deaths (Table II) were recorded between 1960–97, yielding an SMR of 0.64 (95% CI = 0.61–0.67). Overall cancer mortality was also lower than in the general population (SMR = 0.68; 95% CI = 0.63–0.74). Mortality from malignant melanoma was increased significantly (SMR = 1.78, 95% CI = 1.15-2.67), and lung cancer mortality was very low (SMR = 0.53, 95% CI = 0.44–0.62). For most other cancers, the mortality of cockpit crew was slightly lower than in the general population, although the SMR differences were not significant statistically. Thirty pilots died from leukemia, suggesting no excess risk (SMR = 1.05, 95% CI = 0.69–1.50). The results were similar for leukemia excluding chronic lymphatic leukemia (non-CLL leukemia; SMR = 1.12, 95% CI = 0.67–1.70). There were 41 brain cancer deaths yielding an SMR of 1.20 (95% CI = 0.87–1.67). Among rare cancer types not listed in Table II, 3 cases of cancer of the eye were observed vs. about 1 case expected (SMR = 2.93, 95% CI = 0.58–8.42).

Table II. Observed and Expected Number of Deaths, SMR1 and 95% CI Among Male Cockpit Crew in 9 European Countries (ESCAPE Study, 1960–1997)
Cause of death2ICD 9 codeObserved number of deaths (O)Calculated number of deaths4 (Oc)Expected number of deaths (E)SMR (Oc/E)95% CIp for heterogeneity
  • 1

    Standardized mortality ratio.

  • 2

    Non-cancer causes: only diagnoses with 20 or more observed or expected cases; for cancer: 5 or more observed or expected cases.

  • 3

    Excludes cohorts from Germany and Greece as no subtype of leukemia available.

  • 4

    Calculated according to Rittgen and Becker.25

All causes001–999224422443508.50.640.61–0.670.00
Infectious diseases001–1391010.725.70.410.20–0.780.49
Influenza and pneumonia480–4872525.977.30.340.22–0.500.07
All cancer140–208677716.61053.10.680.63–0.740.00
 Buccal cavity/pharynx140–1491313.725.30.540.29–0.950.00
 Oesophagus1501919.834.00.580.35–0.920.08
 Stomach1513436.476.00.480.33–0.670.00
 Colon1536467.963.71.070.81–1.380.87
 Rectum1543031.839.70.800.54–1.160.51
 Biliary tree/liver155–1562323.927.80.860.55–1.330.18
 Pancreas1573638.545.40.850.58–1.180.70
 Larynx16188.613.50.630.27–1.260.96
 Lung162–163153162.2307.70.530.44–0.620.00
 Malignant melanoma1722526.414.81.781.15–2.670.44
 Prostate1855456.760.10.940.71–1.260.83
 Testis/other male genital organ186–18744.27.70.550.15–1.410.89
 Kidney189.0–21515.826.70.590.33–1.000.99
 Bladder/other urinary tract188, 189.3–4, 189.8–91920.534.20.600.35–0.930.59
 CNS191–1924142.935.71.200.87–1.670.39
 Thyroid gland/other endocrine193–19455.43.61.480.47–3.480.02
 All Lymphoma200–2022526.435.20.750.48–1.130.92
  Non Hodgkin's Lymphoma200/202181926.60.710.42–1.150.99
  Hodgkin's Lymphoma20177.48.70.860.34–1.800.78
 All leukemia204–2083032.130.61.050.69–1.500.57
  Leukemia non CLL3204–208, not 204.12122.620.21.120.67–1.700.14
Diabetes2501414.742.80.340.19–0.590.98
Cerebrovascular disease430–438109114.7227.10.510.41–0.620.56
All cardiovascular390–429534565.91118.10.510.46–0.560.00
Acute myocardial infarction410301317.8635.90.500.44–0.570.00
Non-malignant respiratory disease460–479, 488–5193639.4163.10.240.16–0.330.00
Liver cirrhosis5714548.186.60.560.40–0.750.00
Motor vehicle accidentsE810–E8256265.894.70.700.53–0.900.49
Aircraft accidentsE840–E844244258.32.987.7376.5–100.80.00
SuicidesE950–E9597379.1125.80.630.48–0.790.00
All other external causesRest E800–E9998185.9117.00.730.58–0.920.04
All external causes except aircraftE800–E999, not E840–E844236251.7367.90.680.59–0.780.00
Ill defined cause of death780–799.92021.660.70.360.21–0.550.94

Mortality from chronic diseases such as diabetes, respiratory diseases and cardiovascular causes was markedly lower than in the reference population (Table II). The observed number of cerebrovascular deaths was only half the expected (SMR = 0.51; 95% CI 0.41–0.62). More than 10% of all recorded deaths (n = 244) among cockpit crew were due to aircraft accidents (SMR = 88, 95% CI = 77–101), a rare cause of death in the general population. Mortality from other accidents (e.g., motor vehicle) and other external causes was decreased significantly.

Analyses according to time since first employment (data not shown) showed highest SMRs for all causes during the first 10 years after employment and there was no SMR increase with increasing time. For cardiovascular diseases a moderate mortality increase over time was obvious, but the mortality remained at about 50% of the population rates even in the highest category. The Poisson regression analysis to evaluate the influence of duration of employment/license as a proxy for occupational exposures (Table III) indicated a decreasing mortality with increasing employment duration for deaths from all causes (p < 0.01), cardiovascular disease (p < 0.01), leukemia (p = 0.02) and aircraft accidents (p < 0.01). For non-CLL leukemia no trend across duration categories was seen. The colon was the only cancer site for which there seemed to be an increased risk with duration of employment (RR = 30+ years vs. 0–10 years = 2.01; 95% CI = 0.82–4.91, p- = 0.06).

Table III. Age- and Calendar Year Adjusted Rate Ratios1 From Poisson Regression For Mortality From Grouped and Individual Causes of Death Among European Male Cockpit Crew, by Duration of Employment (ESCAPE Study, 1960–1997)
Cause of deathDuration of employment4 (years)Rate Ratios95% CIp for trend
  • 1

    Adjustment for country did not change risk coefficients, but created model convergence problems.

  • 2

    Calendar year adjustment in only 2 categories: 1960–<1990 and 1990–1997.

  • 3

    Excludes cohorts from Germany and Greece.

  • 4

    Person-years in the 4 categories are 241,403 (0–<10y), 162,553 (10–<20y), 121,431 (20–<30y), 22,176 (30+y).

All causes0–<10 (ref.)1  
 10–<200.820.72–0.94 
 20–<300.720.63–0.82 
 30+0.660.54–0.81<0.01
All cancer0–< 10 (ref.)1  
 10–<200.800.63–1.01 
 20–<300.840.68–1.03 
 30+0.850.64–1.130.29
Colon0–<10 (ref.)1  
 10–<200.970.52–1.79 
 20–<301.310.76–2.27 
 30+1.650.87–3.140.06
Lung0–<10 (ref.)1  
 10–<200.980.66–1.45 
 20–<301.070.75–1.52 
 30+0.890.57–1.390.90
Malignant melanoma0–<10 (ref.)1  
 10–<200.820.36–1.83 
 20–<300.800.35–1.86 
 30+0.790.23–2.760.65
Prostate20–<10 (ref.)1  
 10–<200.660.39–1.10 
 20–<300.860.56–1.32 
 30+1.000.60–1.680.61
CNS0–<10 (ref.)1  
 10–<201.690.93–3.08 
 20–<301.010.55–1.86 
 30+0.450.16–1.260.10
All Leukemia0–<10 (ref.)1  
 10–<201.020.55–1.86 
 20–<300.300.14–0.64 
 30+0.770.33–1.800.02
non CLL-Leukemia30–<10 (ref.)1  
 10–<201.270.61–2.63 
 20–<300.370.15–0.92 
 30+1.500.60–3.790.53
All cardiovascular0–<10 (ref.)1  
 10–<201.080.83–1.42 
 20–<300.760.59–0.99 
 30+0.550.37–0.81<0.01
Aircraft accidents0–<10 (ref.)1  
 10–<200.580.44–0.75 
 20–<300.580.38–0.86 
 30+1.040.47–2.28<0.01

Country-specific SMR estimates showed marked differences for several causes of death (Table II, Fig. 1a–e). The all cause SMR was 1.43 (95% CI = 1.05–1.90) in Iceland whereas SMR point estimates for all other countries ranged below one. This difference was caused by the particularly high contribution of aircraft accidents to overall deaths in Iceland (60%).

Figure 1.

Standardized mortality ratio (SMR) and 95% CI for grouped and individual causes of death among male cockpit crew, ESCAPE overall and by national cohort. (a) All causes (SMR, 95% CI). (b) All cancer (SMR, 95% CI). (c) All cardiovascular causes (SMR, 95% CI). (d) Lung cancer (SMR, 95% CI). Lower limit of CI = 0 for Greece. (e) Malignant melanoma (SMR, 95% CI). Diamonds represent the SMR estimate; vertical lines the corresponding 95% CI. The horizontal dotted lines are the lower and upper limits of the 95% CI for the pooled SMR; log SMR and log 95% CI displayed on linear scale. ESCAPE, overall cohort; DE, Germany; DK, Denmark; FI, Finland; GR, Greece; IS, Iceland; IT, Italy; NO, Norway; SE, Sweden; UK, United Kingdom.

When aircraft accidents were excluded from the analyses, the overall ESCAPE-SMR was reduced to 0.57, but some heterogeneity was still present, mainly due to a low overall SMR in Germany (SMR = 0.42) and a rather high value for Denmark (SMR = 0.80). Similar differences were seen for all cancers, lung cancer and all cardiovascular deaths. The test for heterogeneity (Table II) was not statistically significant for most other causes of death, but the test power was low in most instances. Malignant melanoma deaths were rare overall, and accordingly the country-specific SMR estimates were imprecise, ranging from 0 (Finland) to 9.1 (Iceland). There was, however, no statistical heterogeneity between countries.

DISCUSSION

Potential health effects of cosmic radiation among cockpit crew have become a subject of concern and research recently. Along with other aircrew they have been classified as occupationally exposed to ionizing radiation since 1996,27 with typical annual exposure levels in the range of 2–6 mSv. This amounts to roughly twice the annual background radiation dose in most countries28 and is greater than the radiation dose that nuclear power plant workers receive. Theoretically a career dose of 150 mSv or more could be possible for a pilot, but detailed dose estimations carried out for some cohorts19, 21 have not indicated cumulative doses exceeding 100 mSv. Nevertheless, late effects of cumulative radiation exposure, most notably excess cancer incidence and mortality, have been of concern.

The current study is the largest mortality study among cockpit crew published so far, covering more than half of all European cockpit crew employed and licensed in the second half of the last century. Consistent with studies both included12, 29 and not9 in this analysis and a strong healthy worker effect in cockpit crew, the all-cause mortality was markedly decreased. The typical “healthy worker pattern” of increasing mortality with increasing time since first employment was not seen, however, mainly because of a high risk of death from aircraft accidents in the early years of employment or licensing. High social status and regular medical examinations may contribute to the low mortality. These reasons may also partly explain the observed low overall cancer mortality and a lower prevalence of life-style factors, particularly smoking, may play an additional role. No data were available on smoking patterns of the overall cohort, but the markedly reduced mortality from lung cancer (SMR = 0.53) suggests that smoking has been less common among pilots and other cockpit members than in the general male population. In addition a cross-sectional study among Canadian pilots30 confirm a very low smoking prevalence of current cockpit crew.

We used the general population of the participating countries as reference population. A better defined and more comparable group would be an alternative. With regard to cockpit crew, air traffic controllers might be such a group, with similar educational level, salary, occupational health requirements and exposure to shift work. An even better option for the future, however, is to focus on internal comparisons if larger numbers of deaths will have occurred after an extended follow-up period.

Both incidence and mortality from malignant melanoma have been increased in previous cockpit crew studies9, 12, 13, 14, 15, 16, 20 most of which are also included in our analysis. It is difficult to demonstrate an excess of melanoma in mortality studies because of the high survival rate for this disease. The 1.8-fold mortality increase among cockpit crew in our study was based on 25 melanoma deaths, with high risks reported mainly from Northern European countries (except Finland). Occupational UV exposure is unlikely as shielding of aircraft windows against ultraviolet radiation is effective.31 More frequent leisure-time sunlight exposure has been suggested as a probable explanation,12 but more information on this issue is clearly required. An association between ionizing radiation and incidence of melanoma remains under debate.32, 33, 34 Supporting a possible melanoma-ionizing radiation link, slight increases of melanoma risk among radiologic technologists were reported recently.35 Among Nordic pilots,20 a positive dose-response relationship between estimated radiation exposure and melanoma incidence was observed although the authors concluded that the excess melanoma incidence may well be attributable to ultraviolet radiation. Possible interactions between ionizing and UV-radiation36 also require further study.

For leukemia, previous studies reported risk estimates ranging from 0.9–1.7 with highest disease risks among Danish jet pilots14 but the case numbers were small in most studies. Based on 30 deaths from leukemia, no differences between mortality rates of cockpit crew and that of the reference population were found. This finding was essentially unchanged when the analysis was restricted to non-CLL leukemia. Consistent with earlier studies37, 38, 39 no increased risks of other radiation-induced cancers were seen. A longer follow-up of our cohorts and analyses using more refined exposure data are needed to further investigate the risk for leukemia and other radiation-associated malignancies in the future.

Increased mortality and incidence has been reported previously for several other malignancies among cockpit crew, including brain, prostate and bowel cancer. The ESCAPE study showed an overall SMR of 1.2 for brain cancer, with a wide confidence interval. There was a tendency for higher risks among those with employment of more than 10 years duration. Occupational risk factors for brain cancer have not been clearly identified.40 In US Air Force pilots, a weak association between brain cancer and electromagnetic fields has been found, but a much stronger association existed with socio-economic status.41 In the Nordic incidence study,20 no increase of brain cancer incidence was observed.

A suggestive dose-response relationship between number of block hours and incidence of prostate cancer among Nordic airline pilots was described recently.20 Our mortality analysis showed no increases for prostate cancer mortality in European cockpit crew, with an overall SMR of 0.94 and no obvious time-related variations in mortality. Differences between incidence and mortality studies could be due to an increased detection rate of early incident prostate cancer among cockpit crew. Occupational exposures (radiation, sedentary work) might be responsible for a small proportion of the excess risk for cancer of the colon associated with longer duration of employment, but other, e.g., dietary factors may also play a role.

Our study confirms the marked cardiovascular mortality reduction of cockpit crew9, 11 as compared to the general population. Obviously only very healthy persons are recruited into the profession, and continuous medical supervision may play a role in controlling cardiovascular risk factors. Analyses of mortality by age show that low mortality persists even after retirement. As seen in ESCAPE subcohorts12, 18, 29 and independent studies8, 9 published previously, aircraft accidents, whether private or occupational cause a substantial proportion of deaths especially among younger cockpit crew. Pilots in early phases of their flight career seem to carry the highest risk of accidental death in aviation. Mortality from motor vehicle and other accidents was low.

Our study has several strengths, among them the large size of the cohort and reliable population registration systems and mortality statistics in most participating countries. The cohort represents a highly complete sample of the source population in the participating countries. Where license data were used, all licensed pilots were included, in the other countries the largest airline companies supplied data. Nevertheless, the results should be interpreted with caution. Limitations of epidemiological studies among aircrew include the low statistical power due to the small expected risk increases particularly for radiation-induced cancers.42 Additionally our cohort is very young. Despite the large number of person-years at follow-up, <10% of cohort members have died and numbers for many specific cancer sites are still small.

Although cohort members across the different participating countries had rather similar working conditions the SMRs for several causes of death were significantly different between national sub-cohorts. Despite the common protocol methodological discrepancies may still partly explain this observation. More important than the slight differences in cohort enumeration, differences in the quality and availability of reliable cause of death data could contribute to the heterogeneity. Differences in exposure, depending on geographical location and routes flown as well as differences in life-style factors and in the staff selection procedure of airlines may also have influenced the mortality risks to a minor extent. Clearly, country-specific risk differences for aircraft accidents contributed to the heterogeneity of the all-cause SMR.

No information on confounding, such as smoking and leisure-time UV exposure, was available. Detailed evaluations of specific causes of death including information on confounders would require additional data collection. It seems currently that an extension of the follow up of the cohort to gain more power for the analyses is feasible and important.

In summary, this large pooled analysis of mortality data from 9 European countries showed a markedly decreased mortality from all causes as well as from cancer and cardiovascular diseases among cockpit crew as compared to the general population. Deaths due to aircraft accidents were relatively frequent among cockpit crew. Mortality from malignant melanoma was raised, but no clear associations with employment patterns emerged. Our results do not indicate a substantial contribution of occupational factors to the mortality from cancer and most other causes among cockpit crew in Europe.

Acknowledgements

The coordination work in this project was supported by a grant from the European Commission (No. BHM4/CT98/3011).

Ancillary