Abstract
There is substantial public concern about the health risks of proximity to petrochemical industries. In the Haifa Bay Area (HBA), which contains Israel’s densest industrial area, these concerns have been strengthened by elevated cancer mortality rates since the late 1960s. We studied the association between adolescent exposure to industrial air pollution in the HBA and adult-onset cancer. This is a historical cohort study. The study population comprised 2,187,317 subjects, using the Israeli medical corps data linked to the Israel National Cancer Registry with follow-up of up to 45 years. Exposure assessments were estimated by a spatial kriging interpolation model of SO2, serving as a marker for the dispersion of air pollution emitted from the complex during the study period. We found increased crude (HR = 1.23, 95%CI= 1.17 to 1.29) and adjusted (HR = 1.16, 95%CI = 1.10 to 1.21) risk of cancer with increased exposure to air pollution in HBA. The associations remained robust in analyses stratified by decade and socio-economic status. We found evidence of monotonically increased risk in five of 13 cancer categories (leukemia, melanoma, female breast, central nervous system, and thyroid tumors). Our findings strengthen the hypothesis that this exposure posed a carcinogenic risk during the study period.
Keywords: cancer, Haifa Bay Area, epidemiology, public health, industrial air pollution
Introduction
The Haifa Bay Area (HBA), located in northern Israel, is one of the country’s largest metropolitan areas. HBA contains the country’s densest industrial area, which includes oil refineries, petrol storage, petrochemical plants, a power plant, a seaport, a small airport, pharmaceutical and chemical factories, metal processing factories, and other industrial facilities. There is substantial public concern about excess health risks posed by the residential proximity to petrochemical industries, particularly regarding cancer. In the HBA, these concerns are further strengthened by elevated age- and sex-adjusted cancer mortality rates compared to other areas in Israel since the late 1960s.1,2
Two comprehensive studies have reviewed associations between proximity to petrochemical industrial areas and adverse health outcomes.3,4 Most published studies have found a higher cancer incidence and mortality, with the leading associated cancers being leukemias5−7 and respiratory tract (mainly lung) cancers.7−10 Other studies found associations with cancer of the liver,11 pancreas,12 brain,13,14 bladder,10,14,15 and various hematological malignancies.10,16−18 Studies that assessed the association with any cancer (regardless of the cancer site) also found positive associations.19−21 However, many of these investigations are limited by low statistical power given the few cases of each specific cancer. Another common limitation is the ecological design, which almost always lacks a detailed exposure assessment and is prone to residual confounding.
Previous epidemiological studies in the HBA have also used an ecological study design, in which both exposures and outcomes were assessed using aggregated measures, lacking individual-level data.22−25 Furthermore, these studies examined associations with criteria pollutants, such as nitrogen oxides or particulate matter, which do not comprise the relevant industrial pollutant exposure in the HBA – and not with volatile organic compounds (VOCs) or metals, which were generally not monitored. One exception to the published ecological studies in the HBA is a pair of two historical cohort studies in adults, which were based on a representative sample of subjects from the Israel Central Bureau of Statistics and assessed individual-level cancer incidence data.26,27 However, these two studies lacked any exposure assessment and cancer risk was merely compared between HBA residents and non-HBA Israelis, without considering the large exposure variability expected among the former.26,27
Disease patterns in the HBA were likely affected by chronic and remote exposures that may no longer exist in the area. Therefore, exposure assessments based on historical measures are required to study factors influencing current, recent, or past morbidity. However, there are three main challenges posed in compiling a historical emission inventory for HBA, particularly of VOCs: 1) most VOCs and toxic metals are rarely monitored even today, and the methods to measure them are technically complex and expensive; 2) simple dispersion models are insufficient to model the complex HBA topography and meteorology; 3) emissions from HBA industry have changed over time, and before the 90s, air pollution monitoring was limited in the area.
The objective of the present study was to examine the association between exposure of adolescents to industrial air pollution in the HBA and adult-onset cancer. First, to address the exposure assessment challenge for the HBA, we developed a model representing the dispersion of industrial air pollution in the area. Then, we approached the epidemiological investigation with individual records on ∼2 million subjects using the Israeli medical corps data linked to the Israel National Cancer Registry with a follow-up of up to 45 years.
Materials and Methods
Study Design and Study Population
The study population for this historical cohort study included all Israeli-born subjects whose medical status was evaluated between 1967 and 2012 when they were 16–20 years old before their compulsory military recruitment, regardless of whether they were recruited or not. Follow-up time started at the examination date and ended at the date of first cancer diagnosis, death, or end of follow-up (31.12.2012), whichever occurred first. We excluded 1,560 adolescents with cancer diagnoses before the examination date, resulting in a study population of 2,187,317 adolescents. Ethical permission for the study with a waiver on individual informed consent was granted by the institutional review board of the Israeli medical corps (protocol 1663-2011).
Case Ascertainment
We linked the medical corps records with cancer incidence up to 2012, obtained from the Israel National Cancer Registry. The registry, established in 1960, meets internationally accepted guidelines for the coding system (ICD-O version 3), with data completeness of 97% for solid tumors and 88% for nonsolid tumors and consistently high coverage.28 We further obtained mortality data up to 2012 from the national death registry.
First, we used an “any cancer” definition composed of all invasive tumors, as well as benign central nervous system (CNS) tumors and cervical intraepithelial neoplasia grade III. Second, we classified these cancer diagnoses into 13 categories by ICD-O topography codes, ICD-O-3 morphology codes, ICD-O-3 behavior codes, and statistical power considerations. These categories included head and neck, gastrointestinal, pulmonary, melanoma, breast (women only), female reproductive organs, male reproductive organs, urinary tract, central nervous system, thyroid, Hodgkin’s lymphoma, non-Hodgkin’s lymphoma (NHL), and leukemia. Detailed cancer sites, topography, morphology, and behavior codes selected for each category are listed in Table S1.
Exposure Data
We estimated the exposure to HBA industrial air pollution (HBA-IAP) using a kriging interpolation model described in detail in a previous publication.29 Briefly, we spatially interpolated the 2002–2004 observed SO2 concentrations in the study area. During the study period, SO2 was emitted almost exclusively by industrial plants in the study area30 and can therefore serve as a marker for the dispersion of the entire mix of HBA-IAP that was coemitted from the industrial complex during that period. During 2002–2004 the number of SO2 monitoring stations peaked, and the SO2 concentrations were still high enough to observe them with a low relative uncertainty. We considered the mean SO2 observations between 2002 and 2004 to represent the time-invariant spatial distribution of HBA-IAP. It was verified that the air pollution dispersion patterns were quite constant during the study period, as the topography-affected seasonal meteorology was rather stable.
The model aims to estimate the HBA-IAP relative spatial variability (not absolute concentration values), enabling the classification of HBA into several spatial exposure categories. We obtained the SO2 concentrations from all the air quality monitoring stations operated by Israel’s Ministry of Environmental Protection, the HBA municipal association for environmental protection, and the Israel Electric Corporation. All the raw exposure data used passed quality assurance procedures.29
The individual exposure assignment was based on geocoding of the residence at the time of the prerecruitment examination. Approximately 90% of the addresses in HBA were geocoded with very good accuracy (full address or street level), and an additional 10% were geocoded at the neighborhood or locality level.
To adjust our associations to other sources of air pollution, we accounted for exposure to nitrogen oxides, measured routinely in the past 20 years (1992–2011) at several air quality monitoring stations. We assessed exposure to nitrogen oxides using a land use regression model with a 50 m resolution described in detail in previous publications.31−33 This model back-extrapolates exposures to nitrogen oxides to our entire study period. For this exposure, we used the average concentration from birth to age 17.
Statistical Methods
We assessed cancer risk using crude and multivariable Cox proportional hazards models, comparing exposure tertiles within the HBA and an additional non-HBA reference category. As a primary dependent variable, we used cancer of any type (“any cancer”), as defined above. The large number of cases for this outcome facilitated subgroup and sensitivity analyses. In addition, we ran basic crude models and adjusted models for each cancer category separately.
Associations were reported as hazard ratios (HR) and 95% confidence intervals (CI) for each HBA-IAP exposure category compared with the non-HBA population. To reduce potential confounding, we adjusted our models for discrete levels of the following covariates: sex, year of birth, locality type (rural, urban <50K residents, or urban >50K residents), parental country of origin, nitrogen oxides, and cognitive score. Of note, the cognitive scores in this cohort were evaluated at the examination date by a general intelligence test; they were found in previous studies to be strong predictors of future health.34−36
In addition, models stratified by birth periods and socioeconomic status explored the effect modification by these factors. Finally, to examine the possibility of residual confounding by coexisting illness, we performed sensitivity analyses, limiting the study sample to participants with unimpaired health at the examination date. This was defined as a lack of morbidity history requiring a chronic medical treatment or follow-up, including any major surgery.37 All analyses were conducted in R, version 3.6.1, with the survival models fitted with the cox.zph function in the survival package.38 We tested the models for violations of the proportionality assumption by varying the interaction terms with time.
Results
The cohort included 2,187,317 participants, of which 59% were males and 9% were residents of the HBA, contributing 41,696,278 person-years of follow-up altogether. Over 45 years, 47,129 participants were diagnosed with cancer, resulting in a crude incidence rate of 142.8 cases per 100,000 person-years in the nonexposed category and 171.5, 171.7, and 174.8 cases per 100,000 person-years in the low, intermediate, and high HBA-IAP exposure levels, respectively (Tables 1a and 1b). The Median age at the end of follow-up was 35 years (interquartile range: 20 years) for the entire cohort and 42 years (interquartile range: 18 years) for cancer patients. The distribution of age at diagnosis by HBA-IAP categories is described in Table S2.
Table 1a. Descriptive Statistics of the Study Population by Cancer Category (N = 2,187,317).
| Characteristic | Entire Cohort | Any Cancer | Head& Neck | GI | Pulmonary | Melanoma | Breast (Female) | Female Reproductive |
|---|---|---|---|---|---|---|---|---|
| Total N | 2,187,317 | 47,129 | 1292 | 4103 | 1925 | 5185 | 8576 | 2038 |
| Sex | ||||||||
| Male | 1,294,570(59.2) | 21,097 (44.8) | 1005 (77.8) | 2764 (67.4) | 1444 (75.0) | 2785 (53.7) | ||
| Female | 892,747 (40.8) | 26,032 (55.2) | 287 (22.2) | 1339 (32.6) | 481 (25.0) | 2400 (46.3) | 8,576 | 2,038 |
| Year of birth | ||||||||
| 1947–1959 | 320,470 (14.7) | 22,154 (47.0) | 782 (60.5) | 2699 (65.8) | 1374 (71.4) | 2549 (49.2) | 4410 (51.4) | 965 (47.4) |
| 1960–1970 | 440,294 (20.1) | 13,990 (29.7) | 329 (25.5) | 985 (24.0) | 430 (22.3) | 1565 (30.2) | 3088 (36.0) | 669 (32.8) |
| 1971–1980 | 571,908 (26.1) | 8269 (17.5) | 120 (9.3) | 363 (8.8) | 100 (5.2) | 860 (16.6) | 985 (11.5) | 322 (15.8) |
| 1981–1996 | 854,646 (39.1) | 2716 (5.8) | 61 (4.7) | 56 (1.4) | 21 (1.1) | 211 (4.1) | 93 (1.1) | 82 (4.0) |
| SES Z score | ||||||||
| first | 425,733 (19.5) | 7810 (16.6) | 315 (24.4) | 836 (20.4) | 459 (23.8) | 531 (10.2) | 1010 (11.8) | 303 (14.9) |
| second | 429,549 (19.6) | 9166 (19.4) | 276 (21.4) | 850 (20.7) | 407 (21.1) | 844 (16.3) | 1562 (18.2) | 389 (19.1) |
| third | 422,347 (19.3) | 8999 (19.1) | 232 (18.0) | 771 (18.8) | 363 (18.9) | 928 (17.9) | 1672 (19.5) | 397 (19.5) |
| fourth | 416,107 (19.0) | 9933 (21.1) | 237 (18.3) | 799 (19.5) | 374 (19.4) | 1226 (23.6) | 1957 (22.8) | 469 (23.0) |
| fifth | 417,808 (19.1) | 10,479 (22.2) | 218 (16.9) | 769 (18.71) | 295 (15.3) | 1562 (30.1) | 2264 (26.4) | 454 (22.3) |
| Missing | 75,774 (3.5) | 742 (1.6) | 14 (1.1) | 78 (1.9) | 27 (1.4) | 94 (1.8) | 111 (1.3) | 26 (1.3) |
| IAP-HBA exposure,N(%) | ||||||||
| Reference | 1,945,991 (90.4) | 41,420 (87.9) | 1145 (88.6) | 3673 (89.5) | 1713 (89.0) | 4429 (85.4) | 7540 (87.9) | 1792 (87.9) |
| 1 | 68,587 (3.2) | 1955 (4.1) | 51 (3.9) | 156 (3.8) | 81 (4.2) | 249 (4.8) | 353 (4.1) | 94 (4.6) |
| 2 | 69,149 (3.2) | 1957 (4.2) | 46 (3.6) | 136 (3.3) | 70 (3.6) | 270(5.2) | 364 (4.2) | 80 (3.9) |
| 3 | 67,871 (3.2) | 1797 (3.8) | 50 (3.9) | 138 (3.4) | 61 (3.2) | 237 (4.6) | 319 (3.7) | 72 (3.5) |
| IAP-HBA exposure, Incidence rate (per 100,000 PY) | ||||||||
| 0 | 142.8 | 3.9 | 12.7 | 5.9 | 15.3 | 26.0 | 6.2 | |
| 1 | 171.5 | 4.5 | 13.7 | 7.1 | 21.9 | 31.0 | 8.3 | |
| 2 | 171.7 | 4.0 | 11.9 | 6.2 | 23.7 | 31.9 | 7.0 | |
| 3 | 174.8 | 4.9 | 13.4 | 5.9 | 23.1 | 31.0 | 7.0 | |
| Cognitive score | ||||||||
| Low | 641,826 (30.2) | 9161 (19.6) | 348 (27.4) | 948 (23.5) | 603 (32.0) | 517 (9.9) | 872 (10.2) | 349 (17.2) |
| Intermediate | 719,878 (33.9) | 11,296 (24.2) | 312 (24.6) | 865 (21.5) | 390 (20.7) | 1074 (20.8) | 1686 (19.7) | 484 (23.8) |
| High | 462,524 (21.8) | 16,919 (36.3) | 430 (33.9) | 1566 (38.9) | 654 (34.8) | 2018 (39.1) | 3830 (44.8) | 846 (41.7) |
| Very high | 300,299 (14.1) | 9266 (19.9) | 179 (14.1) | 651 (16.2) | 235 (12.5) | 1552 (30.1) | 2156 (25.2) | 350 (17.2) |
Table 1b. Descriptive Statistics of the Study Population by Cancer Category (N = 2,187,317).
| Characteristic | Entire Cohort | Male reproductive | Urinary | CNS | Thyroid | Hodgkin’s Lymphoma | NHL | Leukemia |
|---|---|---|---|---|---|---|---|---|
| Total N | 2,187,317 | 2974 | 2325 | 1357 | 2899 | 1872 | 3301 | 1597 |
| Sex | ||||||||
| Male | 1,294,570 (59.2) | 2974 | 1979 (85.1) | 1268 (58.8) | 838 (28.9) | 1066 (56.9) | 2234 (67.7) | 1107 (69.3) |
| Female | 892,747 (40.8) | 346 (14.9) | 890 (41.2) | 2,061 (71.1) | 806 (43.1) | 1,067 (32.2) | 490 (30.7) | |
| Year of birth | ||||||||
| 1947–1959 | 320,470(14.7) | 1613 (54.2) | 1529 (65.8) | 922 (42.7) | 853 (29.4) | 382 (20.4) | 1543 (46.7) | 779 (48.8) |
| 1960–1970 | 440,294 (20.1) | 569 (19.1) | 576 (24.8) | 618 (28.6) | 987 (34.0) | 537 (28.7) | 1010 (30.6) | 430 (26.9) |
| 1971–1980 | 571,908 (26.1) | 552 (18.6) | 181 (7.8) | 423 (19.6) | 799 (27.6) | 642 (34.3) | 551 (16.7) | 273 (17.1) |
| 1981–1996 | 854,646 (39.1) | 240 (8.1) | 39 (1.7) | 195 (9.0) | 260 (9.0) | 311 (16.6) | 197 (6.0) | 115 (7.2) |
| SES Z score | ||||||||
| first | 425,733 (19.5) | 507 (17.0) | 460 (19.8) | 429 (19.9) | 438 (15.1) | 361 (19.3) | 665 (20.1) | 342 (21.4) |
| second | 429,549 (19.6) | 568 (19.1) | 468 (20.1) | 420 (19.5) | 599 (20.7) | 366 (19.5) | 675 (20.4) | 319 (20.0) |
| third | 422,347 (19.3) | 615 (20.7) | 472 (20.3) | 410 (19.0) | 591 (20.4) | 351 (18.8) | 603 (18.3) | 302 (18.9) |
| fourth | 416,107 (19.0) | 609 (20.5) | 466 (20.0) | 441 (20.4) | 592 (20.4) | 368 (19.7) | 670 (20.3) | 314 (19.7) |
| fifth | 417,808 (19.1) | 628 (21.1) | 416 (17.9) | 406 (18.8) | 637 (22.0) | 399 (21.3) | 639 (19.4) | 288 (18.0) |
| Missing | 75,774 (3.5) | 47 (1.6) | 43 (1.8) | 52 (2.4) | 42 (1.4) | 27 (1.4) | 49 (1.5) | 32 (2.0) |
| IAP-HBA exposure,N(%) | ||||||||
| Reference | 1,945,991 (90.4) | 2631 (88.5) | 2103 (90.5) | 1887 (87.4) | 2543 (87.7) | 1663 (88.8) | 2942 (89.1) | 1416 (88.7) |
| 1 | 68,587 (3.2) | 108 (3.6) | 86 (3.7) | 89 (4.1) | 98 (3.4) | 76 (4.1) | 140 (4.2) | 59 (3.7) |
| 2 | 69,149 (3.2) | 129 (4.3) | 71 (3.1) | 93 (4.3) | 133 (4.6) | 65 (3.5) | 100 (3.0) | 54 (3.4) |
| 3 | 67,871 (3.2) | 106 (3.6) | 65 (2.8) | 89 (4.1) | 125 (4.3) | 68 (3.6) | 119 (3.6) | 68 (4.3) |
| IAP-HBA exposure, Incidence rate (per 100,000 PY) | ||||||||
| 0 | 9.1 | 7.2 | 6.5 | 8.8 | 5.7 | 10.1 | 4.9 | |
| 1 | 9.5 | 7.6 | 7.8 | 8.6 | 6.7 | 12.3 | 5.2 | |
| 2 | 11.3 | 6.2 | 8.2 | 11.7 | 5.7 | 8.8 | 4.7 | |
| 3 | 10.3 | 6.3 | 8.7 | 12.2 | 6.6 | 11.8 | 6.6 | |
| Cognitive score | ||||||||
| Low | 641,826 (30.2) | 594 (20.2) | 581 (25.3) | 524 (24.8) | 538 (18.7) | 496 (26.8) | 727 (22.3) | 394 (25.4) |
| Intermediate | 719,878 (33.9) | 742 (25.3) | 510 (22.2) | 571 (27.0) | 831 (28.8) | 577 (31.2) | 784 (24.0) | 384 (24.7) |
| High | 462,524 (21.8) | 960 (32.7) | 835 (36.4) | 635 (30.0) | 944 (32.8) | 485 (26.2) | 1152 (35.3) | 529 (34.0) |
| Very high | 300,299 (14.1) | 638 (21.7) | 368 (16.0) | 387 (18.3) | 569 (19.7) | 294 (15.9) | 597 (18.3) | 246 (15.8) |
The most common malignancy detected was female breast cancer, with a crude incidence rate of 26.0 cases per 100,000 person-years in the nonexposed category and 31.0, 31.9, and 31.0 cases per 100,000 person-years in the low, intermediate, and high exposure categories, respectively. The distributions of baseline characteristics for the entire cohort and those with any cancer diagnoses are presented in Tables 1a and 1b, and the distribution of socioeconomic status (SES) quintiles by HBA-IAP exposure categories is described in Table S3.
In unadjusted Cox models, we observed clear, positive exposure–response associations between HBA-IAP levels and any cancer, with an HR of 1.23 (95% CI = 1.17 to 1.29) for the highest exposure category compared with the reference category of non-HBA residency. The association was slightly attenuated when adjusted for potential confounders; HR = 1.16 (95% CI = 1.10 to 1.21) for the highest exposure category (Figure 1, Table S4).
Figure 1.
Associations of exposure to HBA-IAP with any cancer, modeled using Cox proportional hazards regression. 0 = reference category (non-HBA residents), 1 = low HBA-IAP exposure level, 2 = intermediate exposure, 3 = high exposure. Left panel: unadjusted associations. N = 2,151,598, events = 47,129. Right panel: associations adjusted for sex, year of birth, locality type, origin, cognitive score, and NOx. N = 2,022,930, events = 45,141.
When stratifying the study population by the decade of birth, we found monotonic exposure–response curves for the association between exposure and any cancer for all periods, excluding 1981–1996. The magnitude of the association was greater for those born after 1970 (Figure 2, Table 2). In SES-stratified models, we found monotonic exposure–response curves in most SES quintiles, with a stronger association in the lower SES quintile than in all others (Figure 3, Table 3).
Figure 2.
Associations of exposure to HBA-IAP with any cancer, modeled using Cox proportional hazard regression, stratified by the decade of birth and adjusted for sex, locality type, origin, nitrogen oxides, and cognitive score. 1947–1959: N = 278,301, cases = 21,068. 1960–1970: N = 421,408, cases = 13,628. 1971–1980: N = 543,818, cases = 7954. 1981–1996: N = 779,403, cases = 2491. Reference category: non-HBA residents.
Table 2. Associations of Exposure to HBA-IAP with Any Cancer, Modeled Using Cox Proportional Hazard Regression, Stratified by the Decade of Birth and Adjusted for Sex, Locality Type, Origin, Nitrogen Oxides, and Cognitive Score.
|
HBA-IAP
Exposure |
|||
|---|---|---|---|
| Decade of Birth | Low | Intermediate | High |
| 1947–1959 | 1.06 (1.00 to 1.14) | 1.03 (0.96 to 1.10) | 1.10 (1.03 to 1.19) |
| 1960–1970 | 1.05 (0.96 to 1.14) | 1.08 (1.00 to 1.17) | 1.11 (1.01 to 1.21) |
| 1971–1980 | 1.17 (1.04 to 1.31) | 1.15 (1.04 to 1.28) | 1.32 (1.18 to 1.47) |
| 1981–1996 | 1.24 (1.00 to 1.53) | 1.06 (0.85 to 1.31) | 1.33 (1.11 to 1.61) |
Figure 3.
Effect of modification by SES for the association between HBA-IAP and any cancer, modeled using Cox proportional hazard regression, adjusted for sex, year of birth, locality type, origin, cognitive score, and NOx. Higher scores in the census SES index represent higher socioeconomic status. SES 1: N = 425,733, cases = 7522. SES 2: N = 429,549, cases = 8505. SES 3: N = 422,347, cases = 8711. SES 4: N = 416,107, cases = 9631. SES 5: N = 417,808, cases = 10,320. 0 = reference category (non-HBA residents), 1 = low HBA-IAP exposure level, 2 = intermediate exposure, 3 = high exposure. SES = socioeconomic status; HBA-IAP = Haifa Bay Area industrial air pollution.
Table 3. Effect Modification by SES for the Association between HBA-IAP and Any Cancer, Modeled Using Cox Proportional Hazard Regression, Adjusted for Sex, Year of Birth, Locality Type, Origin, Cognitive Score, and NOx.
|
HBA-IAP Exposure |
|||
|---|---|---|---|
| SES | Low | Intermediate | High |
| SES 1 - lowest | 1.04 (0.93 to 1.15) | 1.0 (0.78 to 1.18) | 1.26 (1.06 to 1.51) |
| SES 2 | 1.05 (0.94 to 1.17) | 1.16 (1.05 to 1.28) | 1.14 (1.04 to 1.24) |
| SES 3 | 1.15 (1.03 to 1.28) | 0.96 (0.86 to 1.07) | 1.12 (0.95 to 1.32) |
| SES 4 | 1.04 (0.91 to 1.18) | 1.08 (1.00 to 1.18) | 1.14 (1.06 to 1.24) |
| SES 5 - highest | 1.08 (1.00 to 1.18) | 1.04 (0.95 to 1.14) | 1.10 (0.95 to 1.28) |
Unadjusted Cox models for specific cancers showed monotonic exposure–response associations for several disease groups. The unadjusted HRs in the highest HBA-IAP exposure category compared to the reference category were 1.52 (95% CI = 1.33 to 1.73) for melanoma, 1.36 (95% CI = 1.07 to 1.73) for leukemia, 1.39 (95% CI = 1.16 to 1.66) for thyroid cancer, 1.33 (95% CI = 1.07 to 1.64) for CNS tumors, 1.20 (95% CI = 1.08 to 1.35) for head and neck cancers, and 1.20 (95% CI = 1.08 to 1.35) for female breast cancer (Table 4, left columns).
Table 4. Associations between Exposure to HBA-IAP and Specific Cancersa.
|
HBA-IAP
Exposure |
||||||
|---|---|---|---|---|---|---|
|
Unadjusted |
Adjusted |
|||||
| Cancer type | Low | Intermediate | High | Low | Intermediate | High |
| Head & Neck | 1.06 (0.80 to 1.40) | 0.99 (0.74 to 1.33) | 1.24 (0.94 to 1.64) | 1.05 (0.80 to 1.40) | 0.95 (0.70 to 1.30) | 1.22 (0.91 to 1.63) |
| GI | 0.99 (0.84 to 1.16) | 0.90 (0.76 to 1.10) | 1.07 (0.90 to 1.27) | 0.96 (0.81 to 1.12) | 0.91 (0.76 to 1.08) | 1.03 (0.86 to 1.23) |
| Pulmonary | 1.09 (0.87 to 1.36) | 0.99 (0.78 to 1.26) | 1.00 (0.78 to 1.30) | 1.10 (0.87 to 1.38) | 1.06 (0.82 to 1.35) | 1.01 (0.77 to 1.31) |
| Melanoma | 1.35 (1.19 to 1.53) | 1.51 (1.33 to 1.70) | 1.52 (1.33 to 1.73) | 1.15 (1.01 to 1.30) | 1.17 (1.04 to 1.33) | 1.28 (1.12 to 1.47) |
| Breast (Female) | 1.10 (0.98 to 1.22) | 1.18 (1.06 to 1.31) | 1.20 (1.08 to 1.35) | 1.02 (0.9 to 1.13) | 1.04 (0.93 to 1.16) | 1.14 (1.01 to 1.27) |
| Reproductive (Female) | 1.25 (1.01 to 1.54) | 1.10 (0.88 to 1.37) | 1.14 (0.90 to 1.44) | 1.19 (0.96 to 1.46) | 1.03 (0.82 to 1.29) | 1.12 (0.88 to 1.42) |
| Reproductive (Male) | 0.98 (0.81 to 1.19) | 1.20 (1.01 to 1.43) | 1.14 (0.94 to 1.39) | 0.94 (0.78 to 1.15) | 1.12 (0.93 to 1.33) | 1.12 (0.92 to 1.37) |
| Urinary | 0.95 (0.76 to 1.18) | 0.82 (0.65 to 1.04) | 0.88 (0.69 to 1.13) | 0.96 (0.77 to 1.20) | 0.85 (0.67 to 1.08) | 0.90 (0.70 to 1.16) |
| CNS | 1.15 (0.93 to 1.42) | 1.23 (1.00 to 1.51) | 1.33 (1.07 to 1.64) | 1.16 (0.93 to 1.44) | 1.19 (0.96 to 1.48) | 1.32 (1.06 to 1.64) |
| Thyroid | 0.94 (0.77 to 1.15) | 1.30 (1.09 to 1.55) | 1.39 (1.16 to 1.66) | 0.91 (0.74 to 1.11) | 1.17 (0.98 to 1.40) | 1.28 (1.07 to 1.54) |
| Hodgkin’s Lymphoma | 1.17 (0.93 to 1.48) | 1.00 (0.78 to 1.28) | 1.15 (0.90 to 1.47) | 1.19 (0.94 to 1.50) | 1.00 (0.77 to 1.27) | 1.13 (0.88 to 1.45) |
| NHL | 1.14 (0.96 to 1.35) | 0.84 (0.69 to 1.03) | 1.15 (0.95 to 1.38) | 1.11 (0.9 to 1.32) | 0.84 (0.69 to 1.03) | 1.11 (0.92 to 1.34) |
| Leukemia | 1.01 (0.78 to 1.31) | 0.95 (0.72 to 1.24) | 1.36 (1.07 to 1.73) | 1.01 (0.78 to1.33) | 0.99 (0.75 to 1.30) | 1.37 (1.07 to 1.76) |
Analyses were adjusted by sex, year of birth, locality type, origin, cognitive score, and NOx. HBA-IAP = Haifa Bay Area industrial air pollution; GI = gastrointestinal; CNS = central nervous system, NHL = non-Hodgkin’s lymphoma.
The adjusted associations between HBA-IAP exposure and specific cancers (Table 4, right columns, and Figure 4) demonstrated positive monotonic or nearly monotonic exposure–response curves for five of 13 cancer groups examined: female breast cancer, CNS, leukemia, melanoma, and thyroid cancer. On the other hand, the following cancer categories were not consistently associated with HBA-IAP exposure: head and neck, gastrointestinal tract, Hodgkin’s lymphoma, NHL, male and female reproductive organs, urinary tract, and pulmonary cancer. The findings were comparable in analyses restricted to those with unimpaired health in adolescence (Table S5 and Figure S1).
Figure 4.
Associations between exposure to HBA-IAP and specific cancers were modeled using Cox proportional hazard regression and adjusted for sex, year of birth, locality type, origin, cognitive score, and NOx. Sex was not included as a confounder in the models for female breast cancer and female/male reproductive. N = 2,311,240, N cases: head and neck 1231; gastrointestinal 3909; pulmonary 1821; melanoma 4994; female breast 8315; female reproductive 1952; male reproductive 2830; urinary 2207; CNS 2040; thyroid 2799; Hodgkin’s lymphoma 1792; non-Hodgkin’s lymphoma 3152; lymphoid leukemia 1497. 0 = reference category (non-HBA residents), 1 = low HBA-IAP exposure level, 2 = intermediate exposure, 3 = high exposure. CNS = central nervous system; SES = socioeconomic status; HBA-IAP = Haifa Bay Area industrial air pollution.
Discussion
In this historical cohort study spanning over four decades, we found a positive monotonic exposure–response association between exposure to industrial air pollution in HBA and any cancer, which persists after adjustment for potential confounders. Despite lower precision of the associations expected from the smaller sample sizes, the associations remained robust in analyses stratified by birth decade and SES, and also those restricted to subjects with unimpaired health at baseline.
HBA-IAP likely contains several chemical agents that act by various biological mechanisms, possibly synergistically, to increase the risk of cancer in general and different specific cancers in particular. Our study was not designed to estimate exposure to particular chemicals and thus cannot contribute data on the carcinogenicity of specific agents. Nevertheless, the findings support a generalized enhanced cancer risk with increased exposure to HBA pollutants. This justifies public concern and provides evidence of increased risk in five of 13 specific cancer categories for which monotonic associations are evident (leukemia, melanoma, female breast, CNS, and thyroid tumors). Still, we found inconsistent or null findings in other cancer groups.
Notably, the list of cancer categories that presented clear associations with HBA-IAP includes endocrine-related tumors (thyroid and breast), suggesting the involvement of endocrine disruptors in the unknown mix of HBA-IAP exposures. This is plausible, considering the mixed VOC emissions expected from a petrochemical industrial area. Melanoma risk is elevated in petrochemical workers,39 and although family history, skin type, and sun exposure are the main risk factors for this disease, its incidence may also be potentially influenced by hormones. It is noteworthy that breast cancer, thyroid cancer, and melanoma are amenable to early detection at preclinical stages, raising the possibility that the observed associations were influenced by incomplete control of SES, education, screening, and other health behaviors, or family history of the disease. On the other hand, gastrointestinal tumors, mainly colon cancer, which are also amenable to early detection, were not associated with HBA-IAP, although our cohort may have been too young to detect such an association.
Notably, some cancers typically associated with ambient air pollution, mainly lung, bladder, and prostate, were not associated with HBA-IAP in our cohort. These null associations may be attributed to the more common occurrence of these diseases at older ages, while most of the follow-up in our cohort occurred in young and middle-aged adults who had reached a median age of 35 at the end of follow-up. With that respect, three of the four cancer categories presenting monotonic association with HBA-IAP (female breast cancer, thyroid, and CNS) were implicated to have a disproportionately higher incidence among young adults.40 While the causes for this trend are not fully understood, air pollution was explicitly pointed out as a less studied exposure. Of note, the associations of leukemia and CNS tumors with HBA-IAP are consistent with prior epidemiological studies, where exposure was defined by residential proximity to petrochemical plants.3 Moreover, the literature regarding the role of solvents, including benzene and other VOCs, in the etiology of these tumors is consistent.3 DNA damage, as measured by the comet assay, has been observed even at low levels of benzene exposure in individuals exposed to petrochemicals.41 While residential proximity to petrochemical facilities has been associated with childhood leukemia,42 the present study also provides evidence of an association with adult disease.
Interestingly, we found a stronger adjusted association between HBA-IAP and the risk of any cancer in the lowest SES quintile. This is consistent with other examples of effect modification of environmental factors by SES, such as in the cases of ambient temperature and nonaccidental mortality43 or ozone and childhood respiratory disease.44 In both of these examples and many others, lower SES populations are found to be more vulnerable to environmental effects.45 The reasons for this phenomenon may include interactions of the environmental agent with additional exposures (such as diet and physical activity) and reduced use of preventive health services due to limited accessibility or awareness.
Our study has several strengths. First, our exposure model enabled us to examine exposure–response curves, which provide more robust support for a causal relation than previous HBA studies. Second, the individual-level data allowed us to utilize a cohort design instead of the commonly used ecological method, which is much more prone to confounding. And third, the sample size provided sufficient statistical power to examine specific cancers and carry out stratified analyses, enabling the exploration of effect modification.
Nevertheless, our study has several limitations. We lacked information on major risk factors, i.e., adult weight change and obesity, occupation, familial history, and utilization of screening tests. However, we adjusted our models for sex, year of birth, locality type, parental country of origin, nitrogen oxides, and cognitive scores, reducing possible residual confounding. Furthermore, we had information on adolescent body mass index (BMI), which tracks well into adulthood.46 In addition, we acknowledge the lack of individual-level or area-level data with relevant spatial resolution on smoking. However, one should note that our study included adolescents who were examined over a range of 45 years. While smoking rates somewhat varied over this period,47 the consistency of the results from models stratified by decades did not suggest that the findings are confounded by factors that changed over time. Moreover, had the pattern of associations been a result of residual confounding by smoking, we would expect to see an increased risk of lung cancer with increasing exposure, which is not the case. Therefore, although we cannot rule out confounding by smoking, several arguments do not support this possibility for our positive findings, and this mainly concerns the null association with lung cancer in our study.
Another limitation is the lack of good characterization or a direct measurement of HBA-IAP’s specific chemical and physical makeup. Quantities of particular VOC emitted in HBA are not reported, and these substances’ transformations and reactions in the atmosphere are unknown. Prior attempts to characterize HBA exposure were based on either distance from the industrial area, measurement of criteria pollutants, or simply comparing HBA to non-HBA residents.22−27 We used an alternative approach to categorize the HBA population by levels of exposure to HBA-IAP, enabling us to examine exposure–response relations, an important component for causal inference.
Moreover, the exposure in our study aims to represent the real-world setting in which individuals are exposed to various pollutants, which has profound relevance to public health in the area.
We also lacked residential address data before and after age 17 and school addresses, which may account for a substantial portion of daily childhood and adolescent exposures. Kipnis, in a study of residential mobility in the 1980s, showed that HBA residents were likely to move from houses in low to higher-income areas.48 However, each of our exposure categories contained both high- and low-income neighborhoods, suggesting that such mobility would present minimal bias due to exposure misclassification. In addition, we limited our sample to Israeli-born adolescents to minimize this problem but acknowledged that we could not differentiate between subjects for whom the exposure persisted and subjects for whom it ceased.
We also acknowledge that our findings may not represent the associations in question for the entire population living in the HBA. First, we excluded childhood cancer cases to establish the temporality principle of cohort studies by which the exposure must precede the outcome. Establishing this principle is aimed at giving additional support for a causal interpretation by avoiding reverse causality, but the exclusion of childhood cases probably reduces the precision of the estimates for cancers common in childhood. Second, the cohort is limited to adolescents who were invited for assessment toward military recruitment. Approximately 20% of the population living in Israel belongs to the Arab minority, and most of this group is not invited to the evaluation in the military corps and does not serve in the military. However, a recent study on the association between HBA residence and cancer incidence in adults found a stronger association in the Arab population compared to the Jewish group,27 implying that our findings probably underestimate the association in the entire population.
Conclusion
The finding of increased crude and adjusted generalized risk of cancer with increased exposure to HBA-IAP strengthens the hypothesis that these exposures posed a carcinogenic risk during the study period. This conclusion has major public health implications for the area itself, as the industrial zone in the HBA is located in a highly populated area, and for other populations exposed to similar industries worldwide.
Acknowledgments
We wish to acknowledge the role of Professor Jeremy D. Kark, a colleague and a mentor for all of us, who set the basis for this research and passed away before the completion of this work.
Data Availability Statement
Patients’ data used for this study cannot be shared due to IRB restrictions.
Supporting Information Available
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/envhealth.3c00030.
Table S1. Cancer Site Groups and corresponding codes. Table S2. Distribution of age at diagnosis by HBA-IAP categories. Table S3. Distribution of SES quintiles by HBA-IAP categories. Table S4. Unadjusted and adjusted associations between exposure to HBA-IAP and any cancer. Table S5. Adjusted associations between HBA-IAP and cancer in subjects with unimpaired health at adolescence. Figure S1. Associations of exposure to HBA-IAP with specific cancers, modeled using Cox proportional hazard regression, adjusted for sex, year of birth, locality type, origin, cognitive score, and NOx, in participants with unimpaired health at adolescence (PDF)
Author Contributions
Conceptualization R.R., G.T., O.P., and D.M.B.; methodology, R.R., Y. and D.M.B; validation, R.L.B.-O., D.T., E.D., and R.R.; formal analysis, R.R. and R.L.B.-O.; resources, R.R., D.M.B., G.T., and D.T.; data curation, R.R., R.L.B.-O., D.T., and E.D.; investigation, R.R., and R.L.B.-O.; writing—original draft preparation, R.L.B.-O., and R.R.; writing—review and editing, D.M.B., L.K.-B., R.S., O.P., Y. and G.T.; visualization, R.R., and R.L.B.-O; supervision, R.R. and G.T.; project administration, R.R., G.T., R.S. and D.M.B. All authors have read and agreed to the published version of the manuscript.
The Israel Ministry of Environmental Protection funded this research, grant number 161-2-9.
Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of the Israeli medical corps (protocol code 1663-2011).
Informed Consent Statement: Patient consent was waived by the IRB.
The authors declare no competing financial interest.
Supplementary Material
References
- Ginsberg G. M. Standardized Mortality Ratios for Israel, 1969–78. Isr J. Med. Sci. 1983, 19 (7), 638–643. [PubMed] [Google Scholar]
- Ginsberg G. M. Standardized Mortality Ratios for Israel, 1983–86. Isr J. Med. Sci. 1992, 28 (12), 868–877. [PubMed] [Google Scholar]
- Domingo J. L.; Marquès M.; Nadal M.; Schuhmacher M. Health Risks for the Population Living near Petrochemical Industrial Complexes. 1. Cancer Risks: A Review of the Scientific Literature. Environ. Res. 2020, 186, 109495 10.1016/j.envres.2020.109495. [DOI] [PubMed] [Google Scholar]
- Marquès M.; Domingo J. L.; Nadal M.; Schuhmacher M. Health Risks for the Population Living near Petrochemical Industrial Complexes. 2. Adverse Health Outcomes Other than Cancer. Sci. Total Environ. 2020, 730, 139122 10.1016/j.scitotenv.2020.139122. [DOI] [PubMed] [Google Scholar]
- Weng H.-H.; Tsai S.-S.; Chiu H.-F.; Wu T.-N.; Yang C.-Y. Association of Childhood Leukemia with Residential Exposure to Petrochemical Air Pollution in Taiwan. Inhal Toxicol 2008, 20 (1), 31–36. 10.1080/08958370701758734. [DOI] [PubMed] [Google Scholar]
- García-Pérez J.; López-Abente G.; Gómez-Barroso D.; Morales-Piga A.; Pardo Romaguera E.; Tamayo I.; Fernández-Navarro P.; Ramis R. Childhood Leukemia and Residential Proximity to Industrial and Urban Sites. Environ. Res. 2015, 140, 542–553. 10.1016/j.envres.2015.05.014. [DOI] [PubMed] [Google Scholar]
- Yang C. Y.; Cheng M. F.; Chiu J. F.; Tsai S. S. Female Lung Cancer and Petrochemical Air Pollution in Taiwan. Arch. Environ. Health 1999, 54 (3), 180. 10.1080/00039899909602257. [DOI] [PubMed] [Google Scholar]
- Gottlieb M. S.; Shear C. L.; Seale D. B. Lung Cancer Mortality and Residential Proximity to Industry. Environ. Health Perspect 1982, 45, 157–164. 10.1289/ehp.8245157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- López-Abente G.; Fernández-Navarro P.; Boldo E.; Ramis R.; García-Pérez J. Industrial Pollution and Pleural Cancer Mortality in Spain. Sci. Total Environ. 2012, 424, 57–62. 10.1016/j.scitotenv.2012.02.047. [DOI] [PubMed] [Google Scholar]
- Belli S.; Benedetti M.; Comba P.; Lagravinese D.; Martucci V.; Martuzzi M.; Morleo D.; Trinca S.; Viviano G. Case-Control Study on Cancer Risk Associated to Residence in the Neighbourhood of a Petrochemical Plant. Eur. J. Epidemiol 2003, 19 (1), 49–54. 10.1023/B:EJEP.0000013395.38701.c2. [DOI] [PubMed] [Google Scholar]
- Yang C. Y.; Chiu H. F.; Chiu J. F.; Kao W. Y.; Tsai S. S.; Lan S. J. Cancer Mortality and Residence near Petrochemical Industries in Taiwan. J. Toxicol Environ. Health 1997, 50 (3), 265–274. 10.1080/009841097160474. [DOI] [PubMed] [Google Scholar]
- Pickle L. W.; Gottlieb M. S. Pancreatic Cancer Mortality in Louisiana. Am. J. Public Health 1980, 70 (3), 256–259. 10.2105/AJPH.70.3.256. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Liu C.-C.; Chen C.-C.; Wu T.-N.; Yang C.-Y. Association of Brain Cancer with Residential Exposure to Petrochemical Air Pollution in Taiwan. J. Toxicol Environ. Health A 2008, 71 (5), 310–314. 10.1080/15287390701738491. [DOI] [PubMed] [Google Scholar]
- Pan B. J.; Hong Y. J.; Chang G. C.; Wang M. T.; Cinkotai F. F.; Ko Y. C. Excess Cancer Mortality among Children and Adolescents in Residential Districts Polluted by Petrochemical Manufacturing Plants in Taiwan. J. Toxicol Environ. Health 1994, 43 (1), 117–129. 10.1080/15287399409531908. [DOI] [PubMed] [Google Scholar]
- Tsai S. S.; Tiao M. M.; Kuo H. W.; Wu T. N.; Yang C. Y. Association of Bladder Cancer with Residential Exposure to Petrochemical Air Pollutant Emissions in Taiwan. J. Toxicol Environ. Health A 2008, 72 (2), 53–59. 10.1080/15287390802476934. [DOI] [PubMed] [Google Scholar]
- Ramis R.; Diggle P.; Boldo E.; Garcia-Perez J.; Fernandez-Navarro P.; Lopez-Abente G. Analysis of Matched Geographical Areas to Study Potential Links between Environmental Exposure to Oil Refineries and Non-Hodgkin Lymphoma Mortality in Spain. Int. J. Health Geogr 2012, 11, 4. 10.1186/1476-072X-11-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wilkinson P.; Thakrar B.; Walls P.; Landon M.; Falconer S.; Grundy C.; Elliott P. Lymphohaematopoietic Malignancy around All Industrial Complexes That Include Major Oil Refineries in Great Britain. Occup Environ. Med. 1999, 56, 577–580. 10.1136/oem.56.9.577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Di Salvo F.; Meneghini E.; Vieira V.; Baili P.; Mariottini M.; Baldini M.; Micheli A.; Sant M. Spatial Variation in Mortality Risk for Hematological Malignancies near a Petrochemical Refinery: A Population-Based Case-Control Study. Environ. Res. 2015, 140, 641–648. 10.1016/j.envres.2015.05.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yuan T.-H.; Shen Y.-C.; Shie R.-H.; Hung S.-H.; Chen C.-F.; Chan C.-C. Increased Cancers among Residents Living in the Neighborhood of a Petrochemical Complex: A 12-Year Retrospective Cohort Study. Int. J. Hyg Environ. Health 2018, 221 (2), 308–314. 10.1016/j.ijheh.2017.12.004. [DOI] [PubMed] [Google Scholar]
- Sans S.; Elliott P.; Kleinschmidt I.; Shaddick G.; Pattenden S.; Walls P.; Grundy C.; Dolk H.; Elliott P.; Area S. Cancer Incidence and Mortality near the Baglan Bay Petrochemical Works, South Wales. Occup Environ. Med. 1995, 52, 217–224. 10.1136/oem.52.4.217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Salerno C.; Berchialla P.; Palin L. A.; Vanhaecht K.; Panella M. Cancer Morbidity of Residents Living near an Oil Refinery Plant in North-West Italy. Int. J. Environ. Health Res. 2013, 23 (4), 342–351. 10.1080/09603123.2012.733938. [DOI] [PubMed] [Google Scholar]
- Paz S.; Linn S.; Portnov B. A.; Lazimi A.; Futerman B.; Barchana M. Non-Hodgkin Lymphoma (NHL) Linkage with Residence near Heavy Roads-A Case Study from Haifa Bay, Israel. Health Place 2009, 15 (2), 636–641. 10.1016/j.healthplace.2008.10.004. [DOI] [PubMed] [Google Scholar]
- Zusman M.; Dubnov J.; Barchana M.; Portnov B. A. Residential Proximity to Petroleum Storage Tanks and Associated Cancer Risks: Double Kernel Density Approach vs. Zonal Estimates. Sci. Total Environ. 2012, 441, 265–276. 10.1016/j.scitotenv.2012.09.054. [DOI] [PubMed] [Google Scholar]
- Eitan O.; Yuval; Barchana M.; Dubnov J.; Linn S.; Carmel Y.; Broday D. M. Spatial Analysis of Air Pollution and Cancer Incidence Rates in Haifa Bay, Israel. Sci. Total Environ. 2010, 408 (20), 4429–4439. 10.1016/j.scitotenv.2010.06.031. [DOI] [PubMed] [Google Scholar]
- Portnov B. A.; Dubnov J.; Barchana M. Studying the Association between Air Pollution and Lung Cancer Incidence in a Large Metropolitan Area Using a Kernel Density Function. Socioecon Plann Sci. 2009, 43 (3), 141–150. 10.1016/j.seps.2008.09.001. [DOI] [Google Scholar]
- Rottenberg Y.; Zick A.; Barchana M.; Peretz T. Organ Specific Cancer Incidence in an Industrial Sub-District : A Population-Based Study with 12 Years Follow Up. Am. J. Cancer Epidemiol. Prevention 2013, 1 (1), 1–10. [Google Scholar]
- Kayyal-Tarabeia I.; Blank M.; Zick A.; Agay-Shay K. Residence near Industrial Complex and Cancer Incidence: A Registry-Based Cohort of 1,022,637 Participants with a Follow-up of 21 Years, Israel. Environ. Res. 2023, 216, 114471. 10.1016/j.envres.2022.114471. [DOI] [PubMed] [Google Scholar]
- Moore E.; Silverman B. G.; Fishier Y.; Ben-Adiva E.; Davidov O.; Dichtiar R.; Edri H.; Zatlawi M.; Keinan-Boker L. An Assessment of the Completeness and Timeliness of the Israel National Cancer Registry. Isr. Med. Assoc J. 2021, 23 (1), 23–27. [PubMed] [Google Scholar]
- Raz R.; Yuval; Lev Bar-Or R.; Kark J. D.; Sinnreich R.; Broday D. M.; Harari-Kremer R.; Bentur L.; Gileles-Hillel A.; Keinan-Boker L.; Lyubarsky A.; Tsur D.; Afek A.; Levin N.; Derazne E.; Twig G. Associations between Exposure to Industrial Air Pollution and Prevalence of Asthma and Atopic Diseases in Haifa Bay Area. Atmosphere (Basel) 2021, 12 (4), 516. 10.3390/atmos12040516. [DOI] [Google Scholar]
- The 2001 Annual Report of the Haifa District Municipalities for the Environment (Hebrew); Haifa District Municipalities for the Environment, 2001.
- Levy I.; Levin N.; Yuval; Schwartz J. D.; Kark J. D. Back-Extrapolating a Land Use Regression Model for Estimating Past Exposures to Traffic-Related Air Pollution. Environ. Sci. Technol. 2015, 49, 3603. 10.1021/es505707e. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cohen G.; Levy I.; Yuval; Kark J. D.; Levin N.; Broday D. M.; Steinberg D. M.; Gerber Y. Long-Term Exposure to Traffic-Related Air Pollution and Cancer among Survivors of Myocardial Infarction: A 20-Year Follow-up Study. Eur. J. Prev Cardiol 2017, 24 (1), 92–102. 10.1177/2047487316669415. [DOI] [PubMed] [Google Scholar]
- Lev Bar-Or R.; Levy I.; Twig G.; Broday D. M.; Lyubarsky A.; Derazne E.; Tzur D.; Simon B.; Levin N.; Gileles-Hillel A.; Bentur L.; Raz R. Associations of Exposure to Nitrogen Oxides with Prevalent Asthma and Other Atopic Diseases in Israel. Environments 2021, 8 (10), 110. 10.3390/environments8100110. [DOI] [Google Scholar]
- Twig G.; Tirosh A.; Derazne E.; Haklai Z.; Goldberger N.; Afek A.; Gerstein H. C.; Kark J. D.; Cukierman-Yaffe T. Cognitive Function in Adolescence and the Risk for Premature Diabetes and Cardiovascular Mortality in Adulthood. Cardiovasc Diabetol 2018, 17 (1), 154. 10.1186/s12933-018-0798-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lutski M.; Zucker I.; Bardugo A.; Bendor C. D.; Derazne E.; Tzur D.; Novick D.; Raz I.; Pinhas-Hamiel O.; Mosenzon O.; Afek A.; Gerstein H. C.; Twig G.; Cukierman-Yaffe T. Adolescent Cognitive Function and Incident Early-Onset Type 2 Diabetes. EClinicalMedicine 2021, 41, 101138. 10.1016/j.eclinm.2021.101138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Keinan-Boker L.; Levine H.; Leiba A.; Derazne E.; Kark J. D. Adolescent Obesity and Adult Male Breast Cancer in a Cohort of 1,382,093 Men. Int. J. Cancer 2018, 142 (5), 910–918. 10.1002/ijc.31121. [DOI] [PubMed] [Google Scholar]
- Twig G.; Yaniv G.; Levine H.; Leiba A.; Goldberger N.; Derazne E.; Shor D. B. A.; Tzur D.; Afek A.; Shamiss A.; Haklai Z.; Kark J. D. Body-Mass Index in 2.3 Million Adolescents and Cardiovascular Death in Adulthood. N Engl J. Med. 2016, 374 (25), 2430–2440. 10.1056/NEJMoa1503840. [DOI] [PubMed] [Google Scholar]
- Therneau T. M.; Lumley T.. Survival Analysis; [R Package “Survival” Version 3.1–12]. Comprehensive R Archive Network (CRAN) 2020, 2. [Google Scholar]
- Onyije F. M.; Hosseini B.; Togawa K.; Schüz J.; Olsson A. Cancer Incidence and Mortality among Petroleum Industry Workers and Residents Living in Oil Producing Communities: A Systematic Review and Meta-Analysis. Int. J. Environ. Res. Public Health 2021, 18 (8), 4343. 10.3390/ijerph18084343. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ugai T.; Sasamoto N.; Lee H. Y.; Ando M.; Song M.; Tamimi R. M.; Kawachi I.; Campbell P. T.; Giovannucci E. L.; Weiderpass E.; Rebbeck T. R.; Ogino S. Is Early-Onset Cancer an Emerging Global Epidemic? Current Evidence and Future Implications. Nat. Rev. Clin Oncol 2022, 19 (10), 656–673. 10.1038/s41571-022-00672-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fracasso M. E.; Doria D.; Bartolucci G. B.; Carrieri M.; Lovreglio P.; Ballini A.; Soleo L.; Tranfo G.; Manno M. Low Air Levels of Benzene: Correlation between Biomarkers of Exposure and Genotoxic Effects. Toxicol. Lett. 2010, 192 (1), 22–28. 10.1016/j.toxlet.2009.04.028. [DOI] [PubMed] [Google Scholar]
- Onyije F. M.; Hosseini B.; Togawa K.; Schüz J.; Olsson A. Cancer Incidence and Mortality among Petroleum Industry Workers and Residents Living in Oil Producing Communities: A Systematic Review and Meta-Analysis. Int. J. Environ. Res. Public Health 2021, 18 (8), 4343. 10.3390/ijerph18084343. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Huang Z.; Chan E. Y. Y.; Wong C. S.; Zee B. C. Y. Spatiotemporal Relationship between Temperature and Non-Accidental Mortality: Assessing Effect Modification by Socioeconomic Status. Sci. Total Environ. 2022, 836, 155497. 10.1016/j.scitotenv.2022.155497. [DOI] [PubMed] [Google Scholar]
- O’ Lenick C. R.; Chang H. H.; Kramer M. R.; Winquist A.; Mulholland J. A.; Friberg M. D.; Sarnat S. E. Ozone and Childhood Respiratory Disease in Three US Cities: Evaluation of Effect Measure Modification by Neighborhood Socioeconomic Status Using a Bayesian Hierarchical Approach. Environ. Health 2017, 16 (1), 36. 10.1186/s12940-017-0244-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Deguen S.; Amuzu M.; Simoncic V.; Kihal-Talantikite W. Exposome and Social Vulnerability: An Overview of the Literature Review. Int. J. Environ. Res. Public Health 2022, 19, 3534. 10.3390/ijerph19063534. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Furer A.; Afek A.; Sommer A.; Keinan-Boker L.; Derazne E.; Levi Z.; Tzur D.; Tiosano S.; Shina A.; Glick Y.; Kark J. D.; Tirosh A.; Twig G. Adolescent Obesity and Midlife Cancer Risk: A Population-Based Cohort Study of 2·3 Million Adolescents in Israel. Lancet Diabetes Endocrinol 2020, 8, 216. 10.1016/S2213-8587(20)30019-X. [DOI] [PubMed] [Google Scholar]
- Zarka S.; Levine H.; Rozhavski V.; Sela T.; Bar-Ze’ev Y.; Molina-Hazan V.; Rosen L. J. Smoking Behavior Change During Compulsory Military Service in Israel, 1987–2011. Nicotine Tob Res. 2017, 19 (11), 1322–1329. 10.1093/ntr/ntw285. [DOI] [PubMed] [Google Scholar]
- Kipnis B. A. Graph Analysis of Metropolitan Residential Mobility: Methodology and Theoretical Implications. http://dx.doi.org/10.1080/00420988520080251 1985, 22 (2), 179–187. 10.1080/00420988520080251. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Patients’ data used for this study cannot be shared due to IRB restrictions.