Incidence of cancer in the area around Amsterdam Airport Schiphol in 1988–2003: a population-based ecological study

Background Amsterdam Airport Schiphol is a major source of complaints about aircraft noise, safety risks and concerns about long term adverse health effects, including cancer. We investigated whether residents of the area around Schiphol are at higher risk of developing cancer than the general Dutch population. Methods In a population-based study using the regional cancer registry, we estimated the cancer incidence during 1988–2003 in residents of the area surrounding Schiphol. We defined a study area based on aircraft noise contours and 4-digit postal code areas, since historical data on ambient air pollution were not available and recent emission data did not differ from the background urban air quality. Results In residents of the study area 13 207 cancer cases were diagnosed, which was close to the expected number, using national incidence rates as a reference (standardized incidence ratio [SIR] 1.02). We found a statistically significantly increased incidence of hematological malignancies (SIR 1.12, 95% confidence interval [CI]: 1.05, 1.19), mainly due to high rates for non-Hodgkin lymphoma (SIR 1.22, 95% CI: 1.12, 1.33) and acute lymphoblastic leukemia (SIR 1.34, 95% CI: 0.95, 1.83). The incidence of cancer of the respiratory system was statistically significantly decreased (SIR 0.94, 95% CI: 0.90, 0.99), due to the low rate in males (SIR 0.89). In the core zone of the study area, cancer incidence was slightly higher than in the remaining ring zone (rate ratio of the core zone compared to the ring zone 1.05, 95% CI 1.01, 1.10). This was caused by the higher incidence of cancer of the respiratory system, prostate and the female genital organs in the core zone in comparison to the ring zone. Conclusion The overall cancer incidence in the Schiphol area was similar to the national incidence. The moderately increased risk of hematological malignancies could not be explained by higher levels of ambient air pollution in the Schiphol area. This observation warrants further research, for example in a study with focus on substances in urban ambient air pollution, as similar findings were observed in Greater Amsterdam.


Background
Amsterdam Airport Schiphol is one of the main airports of Europe. The airport is a major source of complaints about aircraft noise, noise related adverse health effects andespecially since the crash of an airplane in a suburb of Amsterdam on October 4 th 1992 -about safety risks. A longstanding subject of concern of the surrounding population is the exposure to aviation fuels and their combustion products and an alleged increase of cancer risk. Particularly in warm summers the smell of aviation fuels can be distinguished outside the airport grounds. Aircraft emissions vary with the engine type, the engine load and the kind of fuel. Combustion of aviation fuels results in CO 2 , CO, C e , NO x , particles, and a great number of other organic compounds, among which a number of carcinogens [1]. Among the emitted polycyclic aromatic hydrocarbons no compound characteristic for aircraft engines has been detected so far.
A committee of the Health Council of The Netherlands recently reviewed the data on the health impact of large airports [2]. It was concluded that, generally, integrated health assessments are not available. In the last 30 years, several adverse health effects in relation to exposure to air-Noise exposure (in Kosten-units) in the Schiphol area in 1991 Figure 1 Noise exposure (in Kosten-units) in the Schiphol area in 1991. The area surrounded by the blue line indicates the core zone, the black line includes the total study area. The location of the three air quality monitoring stations are indicated by asterisks (*).
craft noise have been the subject of study, such as the use of tranquillizers, the prevalence of bronchitis and cardiovascular disease as well as child stress responses and cognition [3][4][5][6]. However, little information is available in the international literature on cancer risk in relation to airports.
In the late 1980s, mortality due to cancer in the community of Haarlemmermeer, which hosts Schiphol, was investigated by the Municipal Health Service of Amsterdam on request of the general practitioners in the area [7]. The total cancer mortality and the lung cancer mortality in Haarlemmermeer during 1981-86 did not differ statistically significantly from the cancer mortality in the two standard populations that were used. The mortality due to non-Hodgkin lymphoma (NHL) was statistically significantly increased, but conclusions as to the cause of the excess mortality were not possible.
In the 1990s, we carried out a first study on the incidence of cancer in the vicinity of Schiphol, as part of the health surveillance of the resident population of the Schiphol area [8]. During 1988During -1993, the incidence of cancer in the area around Schiphol was close to the national average. The differences in incidence of certain types of cancer in comparison to the national average, as well as those between two study areas characterized by different levels of increased aircraft noise, were considered to be most likely due to differences in life style, such as smoking. In order to investigate whether cancer risk of the resident population of the Schiphol area (in comparison to the national average) changed since 1988-1993, we contin-ued monitoring cancer incidence and we report here on the second, much larger population-based study of the cancer incidence around Schiphol.

Definition of the study population and the study area
When we designed our first study, relevant exposure data on the ambient air quality around Schiphol airport were lacking and we could not define a study population exposed to increased ambient levels of aircraft emissions. The airport itself has no permanent residents and the most heavily exposed population -the airport personnel and the travelers -cannot be defined geographically. Therefore, we defined our study population as the population most heavily exposed to increased levels of aircraft noise. Since 1994, the ambient air quality outside Schiphol has been monitored and no differences with the background urban air quality have been reported for the compounds that were measured [9]. Table 1 summarizes the results of the three monitoring locations in the Schiphol area. However, it is possible that exposure to aircraft emissions has been greater in the past when aircraft engines used to be technologically and ecologically less advanced. Also, we cannot exclude that certain carcinogenic compounds specific to aviation combustion have not been monitored. Since most cancers have a long induction period and the noise contours are thought to reflect best the historical exposure of the surrounding population to aircraft emissions, we continued to use the levels of aircraft noise to define our study area. The aircraft noise levels of 1991 were available as so-called Kosten-units (Ku) [10]. We used the 35 Ku contour and extended the area with about  1). This total area (surrounded by the solid black line in figure 1) was redefined as 4-digit postal code areas (postal code areas surrounded by grey lines in figure 1). The four airstrips of the airport are easily recognized by noise levels over 50 Ku. We also defined a core zone for the 4-digit postal code areas within the 45 Ku contour (the area bordered by the blue line in figure 1), although we do not have empirical data showing that this zone corresponds to a zone with increased levels of ambient air pollution. The remaining study area surrounding the core zone we designated as 'ring zone'. The location of the three air quality monitoring stations (Badhoevedorp, Hoofddorp, Oude Meer) is also indicated in figure 1. The total study area with a population of 177 000 on 31 December 2003 comprised (parts of) five municipalities (table 2). Table 2 also includes figures on per capita income as approximation for socio-economic status.

Population data
Annual population data covering the period 1995-2003 according to 4-digit postal code, 5-year age groups and sex, were available for all municipalities from Statistics Netherlands. For the period 1988-1994 we used data from the municipal administrations.

Cancer registry data
The Amsterdam Cancer Registry (ACR) is a regional, population-based cancer registry with complete regional coverage since 1988. The ACR is part of the nation-wide Netherlands Cancer Registry (NCR) [11]. Completeness of the NCR is estimated to be over 95%. The information is extracted from the medical records by registration clerks. Apart from demographic data, data are collected on tumor site, morphological classification (according to the International Classification of Diseases for Oncology [ICD-O], versions 1 and 2), stage of the tumor and treatment of the patients. The third version of the ICD-O was introduced in the NCR for cases diagnosed as of January 2001. Cases diagnosed in a hospital outside the ACR region but with residence in the ACR region are routinely obtained from the national registry and included in our regional registry. Consequently, these cases could be included in the study.
We selected from the registry all cancer cases in the period 1988-2003 with residence in the area around Schiphol airport at the date of diagnosis. We stratified the cases according to type of cancer (or group of cancers), area of residence (core zone or the ring zone), 5-year age group and sex.

Statistical methods
In our analysis, the incidence of cancer in the national population of the Netherlands served as the reference entity. The expected numbers of cancer (E) for the Schiphol area were calculated for three periods (1988-1993, 1994-1998 and 1999-2003), based on the population data of the Schiphol area (according to 5-year age category and sex) and the 5-year age category and sex-specific cancer incidence rates from the NCR. For the period 1988-1993 we used the average incidence rates of the   [13]. The expected numbers were compared with the observed numbers (O) and standardized incidence ratios (SIRs) were calculated as the ratio between the observed and expected numbers. Exact 95%-confidence intervals (CI) based on the Poisson distribution of O were calculated using STATA 7.0 (STATA Corporation. College Station, Texas, USA). Rate ratios (RR) for the core zone were calculated by dividing the standardized incidence ratio of the core zone by the rate of the ring zone. Ninety five percent CIs of RRs were calculated assuming a log-normal distribution [14].

Results
In 1988-2003, a total of 13 207 cancers (6 739 in males, 6 468 in females) were diagnosed among residents of the Schiphol area (table 3), which included 2 352 cases among residents of the core zone.  Table 4 shows that cancer incidence in the core zone was slightly increased in comparison to the national incidence (SIR .52) was also relativity high. The incidence of hematological malignancies was higher in the core zone than in the to the ring zone, but the increase was not statistically significant (RR 1.06, 95% CI 0.91, 1.24).

Discussion
The major finding of our study is that total cancer incidence in the area around Schiphol airport was almost equal to the national cancer incidence (SIR 1.02). Furthermore, the incidence of hematological malignancies was statistically significantly increased, while the incidence of cancer of the respiratory system was statistically significantly decreased. We observed an excess risk in children aged 0-14 (SIR 1.26). The cancer incidence in the core zone was slightly increased in comparison to the ring zone, due to an excess risk of cancer of the respiratory tract and prostate in males and cancer of the genital organs in females. ratio of the SIR of the core zone and the SIR of the ring zone CI = confidence interval; SIR = standardized incidence ratio; RR = rate ratio As the overall incidence of cancer of the respiratory tract was decreased (SIR 0.94), this observation does not support a positive association between the airport and the occurrence of cancer of the respiratory tract. The incidence pattern of respiratory system cancer in the Schiphol area, i.e. low rates in males and somewhat higher rates in the core zone and among females, is well within the normal regional variation in the Netherlands. Because smoking is the most important risk factor for lung cancer [15], and there is evidence of substantial regional variation in smoking habits in the Netherlands [16], smoking is likely to be responsible for the differences in respiratory system cancer (mainly lung cancer) between the Schiphol area and the Netherlands overall. Unfortunately, no data on smoking habits according to postal code in the Schiphol area are available. Lung cancer incidence in the 1990s in males was lowest in high income areas in the Netherlands. In females, low rates were found in rural areas, while high rates were observed in urban areas [17]. The slightly increased incidence of cancer of the respiratory system in females is in accordance with the moderately urbanized status of the Schiphol area. The data on per capita income (table 2) support the assumption that the low incidence of cancer of the respiratory system in males is related to the high per capita income of the Schiphol area. However, within the Schiphol area only a weak association was observed between the incidence of lung cancer and per capita income by postal code area (data not shown). This may be due to relatively small numbers by postal code area and the long induction period of lung cancer as the regional variation in lung cancer incidence can best be explained by the smoking habits 10 to 30 years ago.
In a number of studies in urban areas an increase of lung cancer incidence or mortality was observed [18,19], mostly attributed to differences in smoking habits. How-ever, there is increasing evidence for a relation between lung cancer risk and ambient air pollution [20,21]. Although we cannot exclude the possibility that the incidence of cancer of the respiratory system in the absence of the airport would even have been lower than the observed incidence, the pattern of the observed incidence does not render this very likely.
The statistically significantly increased rate for breast cancer in 1994-1998 (SIR 1.09 for the total study area) is also within the observed regional variation in the Netherlands. Part of this variation can be explained by local variation in the start of the national screening program for breast cancer. The relatively high incidence of breast cancer in 1994-1998 is probably related to the start of screening in the Schiphol area in that period.
We do not have an explanation for the relatively high incidence of cancer of the female genital organs in the core zone (RR in comparison to the ring zone 1.24). Possibly, this is only a chance finding, as despite the large variation in risk factors for the specific sites the incidence of all specific sites was increased, while the SIR in comparison to the general population was not statistically significantly increased. An association with pollution has not been described for cancer of the female genital organs. Moreover, in the total study area the incidence of these cancers was not increased (SIR 0.97).
The most striking observation in our study is the increased incidence of hematological malignancies, which was observed consistently over three periods, mostly in males but also in females. The increase was more pronounced in the core zone. The increased incidence was mostly due to increased numbers of cases of ALL and NHL (especially lymphoplasmocytic lymphoma, follicular lymphoma, diffuse large B-cell lymphoma and T-cell lymphoma, but not small lymphocytic lymphoma/chronic lymphocytic leukemia [SLL/CLL]), while the incidence of Hodgkin lymphoma was decreased. However, pathology could not be reviewed in this study and different classification systems of lymphoma have been used by pathologists during the study period.
The moderately increased cancer incidence in children (about one supplementary case per year) was mainly caused by the increased number of cases of ALL, as ALL occurs mainly in children.
Also from a national perspective, the number of cases of NHL was markedly increased. In the Netherlands in 1989-1998, the highest rate of NHL in males was found in Greater-Amsterdam, which includes the Schiphol area (source: the Netherlands Cancer Registry). In females in Greater-Amsterdam, the incidence of NHL was also rela-tively high. Chance is not a likely explanation for our finding of an increased incidence of NHL, since the increase was consistently observed over three time periods. The relatively high incidence of NHL in the Schiphol area is also consistent with an increased mortality due to NHL already reported in Haarlemmermeer in 1981-1986 [7]. In several studies, an increased risk of hematological malignancies was found for farmers, which is related to the use of pesticides, infectious micro-organisms or working with beef cattle [22][23][24][25][26][27]. However, although the Schiphol area includes a few areas with intensive agricultural activities, the increased risk for hematological malignancies was also found in areas with few agricultural activities.
Several studies have shown that the incidence of NHL is correlated with nitrate in municipal drinking water due to nitrogen fertilizers [28,29] and is increased in urban/ industrialized areas [30,31]. Hatzissabas et al found that the incidence of large cell high malignancy lymphomas is highest in industrialized regions with pollution of water supplies by more toxic and immunosuppressive substances, while CLL is more frequent in areas with rather low-dose chronic influences such as from the use of fertilizers and pesticides in farming [32]. The pattern of NHL in the Schiphol area -increase of follicular and diffuse large B-cell lymphoma, but not SLL/CLL -might indicate a relation with pollution which is also found in urban areas.
However, an association between the incidence of hematological malignancies and the environment in the Schiphol area is not supported by the available data on ambient air quality. Measurements in 1989 at the airport grounds of Schiphol showed increased levels of ambient air pollutants, including polycyclic aromatic hydrocarbons which are probably or possibly carcinogenic according to the International Agency for Research on Cancer, but not in the direct vicinity outside the airport ground [33]. Morphology and composition of soot emitted by aircraft at Schiphol showed great similarities with soot emitted by the road traffic. Only different profiles of hydrocarbons in the range of C 6 -C 12 in emissions from aircraft engines, aviation fuels and road traffic were reported. Since 1994, three locations in the vicinity of Schiphol are part of the provincial monitoring network for ambient air quality measurement [34]. During 1994-2002, the concentrations of the air pollutants NO 2 , CO, O 3 , PM 10 (particulate matter <10 µm), benzo(a)pyrene, benzene and black smoke at the three locations in the Schiphol area were stable and well comparable to urban background levels in Amsterdam [9]. A more detailed investigation at 59 additional locations in the Schiphol area in 2000/2001 revealed that the average contribution of air traffic emissions and of aviation fuel storage and transfer to the total concentration of volatile hydrocar-bons in the area around Schiphol were only 3%, and up to 5-7% at individual locations [35]. Road traffic contributed 28%. For CO, NO 2 and PM 10 , no relevant influence of emissions of Schiphol on ambient pollutant levels could be determined. Although we cannot exclude the possibility that residents of the Schiphol area have been exposed to air pollutants that were not measured or that higher levels of air pollutants have existed in the past, the results of the ambient air quality monitoring and the source appointment of air pollutants render it unlikely that aircraft emissions have contributed substantially to the total levels of pollutants in the ambient air of the Schiphol area. It therefore seems unlikely that the increased incidence of hematological malignancies is specifically related to ambient air pollution caused by aircraft emissions.
Our results should be interpreted considering the strengths and limitations of the study design. An advantage is the availability of high quality data from a population-based cancer registry over a period of sixteen years. However, the use of the national cancer incidence as a reference has its limitations. Preferably, the cancer incidence in a population which is comparable to the Schiphol region as far as urbanization, socio-economic status and smoking habits, should be used. Unfortunately, such a reference population is not available. Another limitation of the study is that only cancer cases that were residents of the Schiphol area at the date of diagnosis were included in the study. Part of the original residents will have left the area, while others only recently settled in the area. The effect of migration (non-differential misclassification) usually results in an underestimation of the risk at study.

Conclusion
The overall cancer incidence in the Schiphol area was similar to the national incidence in the Netherlands. An association was found between residence in the Schiphol area and a moderately increased incidence of hematological malignancies, especially NHL and ALL. However, the increased risk of hematological malignancies could not be explained by higher levels of ambient air pollution in the Schiphol area, while similarly increased rates were observed in Greater Amsterdam. Further studies, for example a study with focus on substances in urban ambient air pollution, are necessary in order to elucidate the causes of the observed association.
Publish with Bio Med Central and every scientist can read your work free of charge