Reproductive Toxicology 60 (2016) 104–111
Contents lists available at ScienceDirect
Reproductive Toxicology journal homepage: www.elsevier.com/locate/reprotox
Human sex ratio at birth and residential proximity to nuclear facilities in France Hagen Scherb ∗ , Ralf Kusmierz, Kristina Voigt Helmholtz Zentrum München, German Research Center for Environmental Health, Institute of Computational Biology, Ingolstädter Landstr. 1, D-85764 Neuherberg, Germany
a r t i c l e
i n f o
Article history: Received 13 December 2014 Received in revised form 15 January 2016 Accepted 5 February 2016 Available online 12 February 2016 Keywords: Ecological study Nuclear power plant Nuclear research center Nuclear waste disposal Radiation induced genetic effects Sex odds Trend analysis Uranium mining
a b s t r a c t The possible detrimental genetic impact on humans living in the vicinity of nuclear facilities has been previously studied. We found evidence for an increase in the human secondary sex ratio (sex odds) within distances of up to 35 km from nuclear facilities in Germany and Switzerland. Here, we extend our pilot investigations using new comprehensive data from France. The French data (1968–2011) account for 36,565 municipalities with 16,968,701 male and 16,145,925 female births. The overall sex ratio was 1.0510. Using linear and nonlinear logistic regression models with dummy variables coding for appropriately grouped municipalities, operation time periods, and corresponding spatiotemporal interactions, we consider the association between annual municipality-level birth sex ratios and minimum distances of municipalities from nuclear facilities. Within 35 km from 28 nuclear sites in France, the sex ratio is increased relative to the rest of France with a sex odds ratio (SOR) of 1.0028, (95% CI: 1.0007, 1.0049). The detected association between municipalities’ minimum distances from nuclear facilities and the sex ratio in France corroborates our findings for Germany and Switzerland. © 2016 Elsevier Inc. All rights reserved.
1. Introduction 1.1. Abundance of nuclear facilities Nuclear energy constituted 11% of global electricity production in 2011. Three countries obtain more than half their electricity from nuclear plants (France leads at 78%, followed by Slovakia and Belgium at 54% each), and ten additional countries, all but one in Europe, draw at least 25% from this source [1]. In France, 78% of the country’s electricity is supplied by the 58 currently active nuclear reactors. France is also the largest exporter of nuclear electricity in the European Union and is second only to the United States in terms of total nuclear power production contributing 16 percent to the world’s nuclear-derived electricity. While the environmental and human health risks posed by nuclear power plant accidents are well documented, modeling results by Lelieveld et al. [2] indicate that the occurrence of INES 7 major accidents and the risks of global radioactive contamination have been underestimated. Hence, human exposure risks exist around reactors in densely populated regions, notably in Central Europe and South Asia, where
∗ Corresponding author. E-mail addresses: [email protected] (H. Scherb), [email protected] (R. Kusmierz), [email protected] (K. Voigt). http://dx.doi.org/10.1016/j.reprotox.2016.02.008 0890-6238/© 2016 Elsevier Inc. All rights reserved.
a major reactor accident could subject around 30 million people to radioactive contamination. The recent decision by Germany to phase out its nuclear reactors will reduce the national risk, though a risk by reactors in neighboring countries remains. Furthermore, many nuclear facilities are 30–50 years old, contributing to the potential for catastrophic failure. 1.2. Health risks The possible health risks to populations living near nuclear facilities have prompted studies into the incidence of childhood cancer. For example, a meta-analysis of standardized incidence and mortality rates of childhood leukemia in proximity to nuclear facilities indicated elevated disease rates in the majority of the studies considered, although many findings were not statistically significant [3]. Case-control studies on juvenile cancer and leukemia were performed in Germany [4,5], Switzerland [6], Great Britain [7], and France [8]. Although these studies generally provided limited evidence (possible confounding, restricted statistical power), they nonetheless indicate an increased general human health risk in the vicinity of nuclear facilities. 1.3. Determinants of the sex ratio According to Neel and Schull, the sex ratio, or technically the sex odds, is unique among the genetic indicators [9]. Its uniqueness
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
arises from the fact that maternal exposure is expected to produce sex odds different from sex odds after paternal exposure. This expectation is attributed to the hypothesis that if an Xlinked recessive lethal gene is induced in a mother’s germ cell line by ionizing radiation, it would have no effect on a heterozygous daughter, but would be lethal to a hemizygous male zygote. Xlinked dominant lethal mutations in mothers would be equally lethal to both genders. X-linked dominant mutations induced in fathers would only suppress female offspring. Recessive X-linked lethal mutations in fathers would not influence the gender ratio as sons do not receive the paternal X chromosome and daughters carry (and are protected by) a second X chromosome from their mother [10]. In situations where mothers and fathers are nearly equally exposed on average (e.g. Chernobyl), it seems unlikely or unrealistic that the opposite maternal and paternal effects would precisely cancel each other out. That the paternal effect exceeds the maternal effect seems to be evident from increased overall sex ratios after large scale radiological incidents like the atomic bomb tests, the Windscale fire, and the Chernobyl accident [11–15]. Therefore, the odds of male to female offspring at birth may be a simple and non-invasive way to study and monitor the reproductive status or reproductive health of a population. Scholte and Sobels suggest that the observation of changes in the sex odds in the offspring of irradiated parents may be one of the few methods available for studying the genetic effects of ionizing radiation in humans [16]. Briefly, the survival probability of the female zygote is impacted by a number of lethal factors of varying degree of dominance located on the X chromosome. These factors may be impacted by radiation to either the mother or the father resulting in an impaired X chromosome, thus reducing the viability of female zygotes and changing the sex odds. In accordance with theoretical predictions, Cox found a reduced offspring sex ratio in irradiated women. James, on the other hand, states, “ionizing radiation is the only reproductive hazard, which causes (irradiated) men to sire an excess of sons” [17–19]. Ionizing radiation may differ from other causes of sex ratio variation at birth, because the effect is mediated by direct genetic intervention. It has been hypothesized that other (known) causes of variation of the sex ratio at birth may change parental hormone levels around the time of conception, which affects the probability that fertilization will be by a male sperm, or that altered hormone levels (e.g. by endocrine disrupting chemicals) may be detrimental to the survival of the male embryo [20–23]. It has also been suggested that high androgen levels in fathers (hepatitis B carriers, prostatic cancer patients) may entail male biased offspring [24]. Catalano et al. established that exposure of pregnant women to many forms of stress increases the probability of miscarriage—especially of frail male fetuses [25]. In addition to lethal factors on the X chromosome, Scholte and Sobels allude to nondisjunction resulting in X0 genotypes, which are non-viable in humans and, thus, may also distort the birth sex odds [16]. Down syndrome is a well-known consequence of meiotic nondisjunction in man, and increased Down syndrome prevalence at birth serves to indicate increased nondisjunction across Europe after Chernobyl [26]. More generally, there seems to be an association between maternal radiation history and chromosomally abnormal fetuses, e.g., Klinefelter syndrome (47, XXY) [27]. From the sex determining mechanism in man, involving the X and the Y chromosomes (XX = female, XY = male), it is obvious that the birth sex ratio must necessarily depend on three factors: • ratio of X- and Y-bearing sperm • selection of sperm within the female reproductive tract • differential implantation and survival rates of embryos
105
Investigations in the ratio of X- and Y-bearing sperm did not show significant differences in the X/Y sperm ratio, even in men with three or more same sex children [28]. Recently, Orzack et al. presented data that the sex ratio at birth is due to a higher prenatal female mortality, perhaps due to dysfunction of the paternal X [29]. If the sex ratio near or at conception is 1:1, as the extensive data by Orzack et al. indicate, but the birth sex ratio is male-biased 105:100, then female fetuses are the frailer sex (karyotype) during embryonal and fetal life, possibly mediated through their fathers’ enhanced genetic vulnerability [30]. Findings after Hiroshima and Nagasaki indicate that the sex ratio increase in the offspring of irradiated fathers per unit dose is approximately twice the sex ratio decrease in the offspring of irradiated mothers [10]. In humans, the sex odds at birth is relatively constant at the secular population level, with approximately 105 boys born for every 100 girls [31]. However, considerable variability may be observed under a variety of specific circumstances, including selective abortion in some societies. Many determinants of the sex odds, e.g., race or season as well as methodology to study those determinants have been discussed in the literature [32]. Steiner [33] points out that proposed determinants often showed associations in small samples but could not be replicated in larger populations. This, of course, may be due to insufficient statistical power due to small effects and/or small study-populations. Anthropogenic chemicals and ionizing radiation are determinants of the human secondary sex odds at birth. The effect of mutagenic chemicals or ionizing radiation on sex odds has been well established in animal experiments [34–36]. Stevenson and Bobrow [37] provide a detailed account of methodological issues relevant for the assessment of determinants of the sex odds in humans with special emphasis on the impact of male fetal mortality dynamics on the sex odds. Terrell et al. [38] reviewed approximately 100 publications on possible environmental and occupational determinants of the sex ratio. They concluded that, “limitations in study design and methodological issues make it difficult to draw firm conclusions from the existing sex ratio literature”. This highlights the general difficulties in obtaining firm knowledge about the sex odds determinants in humans. We previously studied the birth sex ratio near nuclear facilities in Germany and Switzerland, and we found evidence for increased gender proportions at birth within distances of up to 40 km from nuclear installations [13,39]. Since many anthropogenic chemicals are also mutagenic, it was natural to employ our spatial temporal methodology [40] to study the possible influence of chemical accidents on the sex odds in the vicinity of chemical plants. We specifically looked at the birth sex odds near Hoechst-Griesheim, the site of an accident in 1993 that spread tons of nitroarenes into the nearby environment [41]. We detected a decrease in the sex odds after the chemical accident [42]. Sociological influences such as stress have also been implicated in sex determination, for example following the earthquakes in Chile [43] and Italy [44]. In accordance with the Trivers-Willard hypothesis [45], these studies suggest a decrease in the human sex odds at birth under adverse living conditions. Lastly, while environmental influences play a significant role, the most dramatic determinant of human sex odds at birth seems to be man-made, namely sex selective abortion. Biased sex ratios pose a problem to societies for example in China and in India [46,47]. 1.4. Objectives Motivated by positive findings of radiation induced genetic effects after the atomic bombing of Japan [10,48], after Windscale/Sellafield [11,49], after the atmospheric nuclear weapon tests [12,13], after Chernobyl [26,50–52], and last but not least in the vicinity of nuclear facilities [39], we decided to study the
106
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
Fig. 1. Position of nuclear facilities in France, see Table 1. Nuclear power plants (grey); nuclear mining, research, storage, and disposal facilities (black).
possible association of the births sex odds with minimum distance from nuclear installations in France. France constitutes a special case in that it has comparatively many nuclear power sites, and thus provides a possibility to repeat the analyses we previously conducted for Germany and Switzerland and potentially confirm our pilot findings [13,39]. Our sex odds studies in Germany and Switzerland comprise all 23 German and Swiss nuclear power plants and 5 other pertinent nuclear facilities. In order to closely approach the methodology of our previous study and to achieve a similar statistical power, we selected all 23 French nuclear power plants complemented by 5 pertinent French nuclear mining, research, storage, and disposal facilities. We thus study 28 major French nuclear facilities in total, analogous to the 28 nuclear facilities previously investigated in Germany and Switzerland. The resulting list of nuclear facilities is presented in Table 1, and their geographical positions in Fig. 1. In Germany and Switzerland, we considered 4,906,027 births in the combined 35 km surroundings [39]. In France the corresponding number of births is 4,219,836, hence a rather similar number. Therefore, our present French study provides a similar statistical precision as our nuclear facility pilot study in Germany and Switzerland.
2. Data and statistical methods 2.1. Births by gender Complete annual French mainland gender specific birth data at the municipality level from 1968 till 2011 were produced by the INSEE “Institut national de la statistique et des études
Table 1 French nuclear power plants and major nuclear facilities. No.
Nuclear power plants
Location
Exposure since
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
BELLEVILLE BLAYAIS BUGEY CATTENOM CHINON CHOOZ CIVAUX CRUAS DAMPIERRE EL4 FESSENHEIM FLAMANVILLE G2/PHENIX GOLFECH GRAVELINES NOGENT PALUEL PENLY RAPSODIE ST. ALBAN ST. LAURENT S/PHENIX TRICASTIN
LENE BRAUD ST. LOUIS ST. VULBAS CATTENOM AVOINE CHARLEVILLE CIVAUX CRUAS DAMPIERRE-EN-BURLY BRENNILIS FESSENHEIM FLAMANVILLE MARCOULE AGEN GRAVELINES NOGENT-SUR-SEINE PALUEL PENLY CADARACHE SAINT-MAURICE-L’EXIL ST. LAURENT DES EAUX CREYS-MALVILLE PIERRELATTE
1986 1980 1971 1985 1962 1966 1996 1982 1979 1966 1976 1984 1958 1989 1979 1986 1983 1989 1966 1984 1968 1985 1979
GRENOBLE CAEN LUXEUIL KRUTH SOULAINES-DHUYS/CSA
1971 1983 1966 1953 1992
Nuclear facilites RESEARCH CENTER 24 RESEARCH CENTER 25 26 STORAGE SITE URANIUM MINING 27 WASTE DISPOSAL 28
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
107
Fig. 2. Jump function (35 km, left) and Rayleigh function (right) for the live births sex odds (male:female) depending on distance from nuclear facilities (NF) in France. The dashed Rayleigh function is the baseline adjusted Rayleigh function in Fig. 4 by Scherb and Voigt [13]; SORbaseline 1.0035 for Germany + Switzerland vs. France.
économiques” (http://www.insee.fr/fr/) and provided by the Centre Maurice Halbwachs (CMH), Partenaire du Reseau Quetelet in Paris, France (http://www.cmh.ens.fr/). The SAS data sets obtained from CHM account for 36,565 different municipalities giving rise to 1,412,109 individual municipality-years with 33,114,626 total, 16,968,701 male, and 16,145,925 female births. The overall sex odds in France from 1968 to 2011 was therefore 1.0510, (95% CI: 1.0502, 1.0517).
nuclear facilities in the German and Swiss studies. For France, we used Lambert-93, a projected orthogonal CRS that is suitable for use in France and Corsica. Lambert-93 is a CRS for large and medium scale topographic mapping and engineering survey, defined using information from IGN—Paris, “Institut national de l’information géographique et forestière” (http://www.ign.fr/). Lambert-93 is well suited for our task of determining the minimum distances of municipalities from nuclear facilities in France.
2.2. Nuclear facilities
2.4. Statistical analyses
The characteristics and geographical positions of the investigated French nuclear facilities including active and inactive nuclear power plants were obtained from the comprehensive documentation by ANDRA, the French “Agence Nationale pour la Gestion des Déchets Radioactifs—National Radioactive Waste Management Agency” (http://www.andra.fr/, http://www.andra.fr/ inventaire2012/#/documents/). Additional information on nuclear power plants, e.g., the year of the “first grid connection”, were obtained from the IAEA “International Atomic Energy Agency” information system (http://www.iaea.org/pris/CountryStatistics/ CountryDetails.aspx?current=FR). In Table 1, we list the 23 French nuclear power plants and the 5 nuclear facilities chosen, their locations, and the first year of possible exposure and effect. Due to the measurement and testing periods necessary before a nuclear power plant can enter commercial use, we set the first year of possible exposure to be one year before the year of the official “first grid connection”. As a rule, each reactor site contains between two and four reactor blocks that became operational over several decades starting in the 1960’s. Fig. 1 presents the geographical positions of the 23 nuclear power plants and the remaining 5 nuclear facilities in France.
We applied linear and nonlinear logistic regression to investigate spatiotemporal trends in the occurrence of boys among all live births in the vicinity of French nuclear facilities, and any potential change in the trends following the activation of those nuclear facilities [53]. We denote the number of male births by m, female births by f, and the fraction of male births as pm = m/(m + f). Two important parameters in this context are the sex odds (SO = pm /(1 − pm ) = m/f), and the sex odds ratio (SOR = SO1 /SO0 = (m1 /f1 )/(m0 /f0 )), which is the ratio of the sex odds between two populations SO1 and SO0, e.g., in exposed versus non-exposed populations. We used dummy coding for single points in time and for time periods as well. For example, the dummy variable for the time window from T onward is defined as dT (t) = 0 for t < T and dT (t) = 1 for t ≥ T. The simple and parsimonious logistic model for a trend and a jump in T has the following form:
2.3. Geo-coding We used the orthogonal Gauss–Krüger coordinate reference system (CRS) for the geographical locations of municipalities and
mt ∼ Binomial((m + f)t ,t ) log odds(t ) = intercept + ˛ × t + ˇ × dT (t) where t denotes the year; mt the number of boys born at time t; (m + f)t the number of live births at time t; and t the probability of a male birth at time t. Completely analogous functions may be defined for distance in place of time. Nonlinear logistic regression for fitting a Rayleigh function [13,39] was carried out by applying nonlinear regression
108
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
Fig. 3. Time trend and jump function for the sex odds within 35 km from the nuclear waste disposal site “Centre de stockage de l’Abue”, jump in the year 2000, jump SOR 1.0717, (95% CI: 1.0101, 1.1370), p-value 0.0218.
to the natural logarithm (log) of the sex odds inversely weighted by the variance of the log sex odds. The data in this study were processed with Microsoft Excel 2010. For statistical analyses, we used R 3.1.0, MATHEMATICA 9, and SAS 9.3 (SAS Institute Inc: SAS/STAT User’s Guide, Version 9.3, Cary NC: SAS Institute Inc., 2012). 3. Results 3.1. Global analyses Fig. 2 (left) presents the live births sex odds (1968 to 2011 combined) in mainland France by the minimum distance of municipalities categorized in 1 km wide distance rings around all 28 pertinent nuclear facilities depicted in Fig. 1. The sex odds ratio SOR between births within and outside of the 35 km distance is 1.0028, (95% CI: 1.0007, 1.0049), Wald Chi-square p-value 0.0079, F-Test pvalue 0.0096, see Fig. 2 (left) bold solid line. We obtain even more precise results when using an impartial Rayleigh function (F-test p-value 0.0018) with 3 parameters instead of only 2 parameters for the simple jump function. The Rayleigh function has the additional advantage of not requiring an arbitrary predefined distance such as 35 km. The estimated baseline sex ratio is 1.0506, (95% CI: 1.0499, 1.0514). The estimated sex odds peaks at 9.1 km, (95% CI: 5.8, 12.4) with a SORpeak 1.0084, (95% CI: 1.0036, 1.0132), see Fig. 2 (right). The jump and Rayleigh function analyses are based on all 33,114,626 births in France from 1968 to 2011. These findings essentially mean that within 35 km from the investigated 28 nuclear facilities in France combined, the sex odds is 1.0535 whereas outside of these combined areas, i.e., in the rest of France, the sex odds is 1.0506. If this increase in the sex odds by the factor 1.0028 (sex odds ratio) in the vicinity of nuclear facilities were exclusively due to a deficit in girls in the denominator, it would correspond theoretically to 5730, (95% CI: 1499, 9982) missing girls in the combined 35 km vicinities of the 28 nuclear facilities. We emphasize that this is probably a conservative quantification of the “real effect” as there are more nuclear facilities in France not yet investigated, and the effect probably ranges farther than 35 km. The latter can be anticipated from inspection of Fig. 2 (left): the 50 km range is also significant (p-value 0.0053). Moreover, we must assume considerable non-differential exposure misclassification biasing our results towards null. Due to the presence of different ionizing materials including neutrons, differing exposure pathways, varying susceptibility of affected people, and relevant nuclear facilities not presently analyzed, it is possible or
even likely that the effect identified in this study (Fig. 2) only partially reflects the presumable “real impact” of nuclear facilities on the secondary sex ratio in the French population. 3.2. France versus Germany and Switzerland For comparison, we included the original Rayleigh curve of our German and Swiss analysis (dashed line in Fig. 2, right) adjusted to account for the different baseline sex odds levels of Germany and Switzerland compared to France [13]. Therefore, the dashed Rayleigh function in Fig. 2 is the baseline adjusted Rayleigh function in Fig. 4 by Scherb and Voigt [13]. The baseline sex odds ratio of Germany and Switzerland compared to France is 1.0035. We previously derived the following parameters from the German and Swiss data: the jump model for distances below 35 km yields a sex odds baseline level of 1.0543, (95% CI: 1.0532, 1.0553) and a sex odds ratio for the jump at 35-km distance of 1.0036, (95% CI: 1.0015, 1.0056), p-value 0.0003. Using an impartial Rayleigh function (pvalue 0.0014), the estimated sex odds peaks at 14.3 km, (95% CI 9.1, 19.5) with a SOR peak = 1.0052, (95% CI 1.0022, 1.0082), p-value 0.0014. Therefore, the French and the German/Swiss 35-km-jump and Rayleigh curve parameters coincide reasonably well within the statistical random variability of the data. 3.3. Central nuclear disposal facility We next took a closer look at French nuclear storage sites (Table 1 and Fig. 1). Here, as in the case of the German interim highly active nuclear waste (HAW) disposal facility Gorleben [54], we found an increase of the human sex odds at birth within 35 km from the “Centre de l’Aube Nuclear Disposal Facility (CSA)” from the year 2000 onward (jump SOR 1.1467, (95% CI: 1.0615, 1.2388), p-value 0.0218. This analysis is based on 75,744 births from 1968 to 2011 in that region (Fig. 3). In order to further investigate the potentially causal relationship between the observed increase in the sex odds around the CSA from 2000 onward, we also investigated the distance dependency of the sex odds ratio (after versus before the year 2000) in 5 km distance rings. In fact, the sex odds ratio peaks within 10–30 km from the site. This effect around the CSA is similar, in shape an in magnitude, to the effect of our previous “Gorleben” analysis [39,54]. Therefore, the question arises whether neutron emitting HAW has also been stored at the CSA by the end of the 1990’s.
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
109
Fig. 4. Spatial trend of the sex odds ratio (SOR: from 2000 vs. before 2000) of aggregated gender specific live births data in 5 km distance rings. The estimated sex odds peaks at 11.6 km, (95% CI: 6.6, 16.6) with an SORpeak 1.1416, (95%CI: 1.0255, 1.2708), F-test p-value 0.0323.
3.4. Luxeuil air base Luxeuil (French: base aérienne Luxeuil) is a medium sized air base located near Luxeuil-les-Bains in the Franche-Comté region in the Northeast of France not far from the German border, see Fig. 1. From May 1966 onward, the air base hosted an operational unit for specialized ammunition storage responsible for the nuclear strike alert facilities on the base. According to the archive “La France nucléaire: matières et sites”, materials processed and stored at Luxeuil include the radioactive elements uranium, tritium, plutonium, and deuterium, see http://web.archive.org/web/20120114121607/http:// www.francenuc.org/fr sites/franche lux f.htm. For this presumed exposure situation, and since the 40 km area around Luxeuil is more densely populated, representing as much as 160,240 births from 1968 to 2011 (compared to 69,534 births around the “Centre de l’Aube” in that same period), we took a closer look at the possible association of the spatial sex odds distribution with distance from Luxeuil. Fig. 5 discloses a distinct and significant increase of the sex odds within 40 km from the nuclear air base. The optimum fit function, in this case, seems to be a (less parsimonious) shifted Gaussian that, contrary to the Rayleigh function, allows for a non-zero sex odds excess at distance zero [53]. However, for proper effect quantification we need a parsimonious approach, as for example a simple jump function at distance 40 km. The jump at 40 km (bold broken line in Fig. 5) yields SORjump 1.0203, (95%CI: 1.0102, 1.0305), p-value < 0.0001. For illustration purposes, we may note that the increased sex odds within 40 km around Luxeuil translates to the potential loss of 1570 girls (=2.03%) per 77,328 girls actually born in that area from 1968 to 2011, given that theoretically only girls and no boys were detrimentally affected by the presumed exposure around Luxeuil. 4. Discussion We found evidence for a significantly elevated human sex odds at birth in the vicinity of 28 nuclear facilities in France, in principle agreement with the results of previous analyses for Germany and Switzerland [13,39]. There is little doubt that mutagenic physical and chemical environmental or occupational hazards can alter the human sex ratio at birth. We observed consistently elevated sex ratios after Chernobyl across Europe, and even as far away as Cuba, contaminated foodstuffs from the former Soviet Union presumably caused increases in the sex ratio [13,15,50,55–57].
Recently, based on our findings, Grech [11,49] has shown that the sex ratio increased in parts of Scandinavia and especially in the most heavily exposed regions of Norway in the ten-year period following the October 1957 Windscale/Sellafield accident in the United Kingdom. Moreover, Grech [12,58] reports a rising birth sex ratio in most regions in temporal association with atomic weapon testing. Since childhood cancer and childhood leukemia are elevated near nuclear power plants, and cancer, birth defects, and sex odds increased after Chernobyl, it was natural to investigate the sex odds in the vicinity of nuclear reactors and more generally near nuclear facilities of all kinds. Based on 20 million births in Germany and Switzerland at the municipality level from 1969 to 2009, we found increased sex odds near nuclear power plants and other nuclear facilities [13,39]. Our present analyses on French nuclear facilities corroborate our previous findings in Germany and Switzerland. In France, we observed a small significant increase of a few per mill in the sex odds around nuclear facilities including all nuclear power plants, and a considerable increase near the waste disposal site CSA in SoulainesDhuys, which is similar in shape and magnitude to the increase around Gorleben, Germany. At the French nuclear air base Luxeuil, we see a long term (1968–2011) highly significant sex odds increase of 2% within 40 km from the site. The strength of our approach is that we analyze total national data and no random samples from data. Therefore, sampling error and sampling confounding is not an issue for our studies. A weakness is of course the highly aggregated nature of our data at an annual municipality basis ignoring alternative sex ratio determinants. However, this obvious drawback is perhaps more than outweighed by the amount of individual births data in the tens of millions, and the corresponding full registration over decades within all municipalities at a national basis. Moreover, it is unlikely that any alternative sex odds determinant (environmental, socioeconomic, ethnic, etc.), other than ionizing radiation, consistently entails spurious sex odds increases near practically all major nuclear facilities. The possibility of incomplete knowledge in the radiation sciences has been emphasized in the nuclear accident context. For instance, unknown exposure pathways or unsuspected radionuclides may pose a threat [59]. The second author of this study has estimated the amounts of thermal neutron activation products near casks containing highly radioactive wastes [60]. The activation product 41-Argon may play a more significant role than previously assumed. Another possibility is that the range and the biological
110
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
Fig. 5. Spatial trend of the sex odds (SO) of aggregated gender specific live births (1968 to 2011) in 5 km distance rings. The smooth fit function (bold solid line) is a shifted Gaussian [53] with p-value 0.0013. The jump at 40 km function (bold broken line) yields SORjump 1.0203, (95%CI: 1.0102, 1.0305), Wald Chi-square p-value
Contents lists available at ScienceDirect
Reproductive Toxicology journal homepage: www.elsevier.com/locate/reprotox
Human sex ratio at birth and residential proximity to nuclear facilities in France Hagen Scherb ∗ , Ralf Kusmierz, Kristina Voigt Helmholtz Zentrum München, German Research Center for Environmental Health, Institute of Computational Biology, Ingolstädter Landstr. 1, D-85764 Neuherberg, Germany
a r t i c l e
i n f o
Article history: Received 13 December 2014 Received in revised form 15 January 2016 Accepted 5 February 2016 Available online 12 February 2016 Keywords: Ecological study Nuclear power plant Nuclear research center Nuclear waste disposal Radiation induced genetic effects Sex odds Trend analysis Uranium mining
a b s t r a c t The possible detrimental genetic impact on humans living in the vicinity of nuclear facilities has been previously studied. We found evidence for an increase in the human secondary sex ratio (sex odds) within distances of up to 35 km from nuclear facilities in Germany and Switzerland. Here, we extend our pilot investigations using new comprehensive data from France. The French data (1968–2011) account for 36,565 municipalities with 16,968,701 male and 16,145,925 female births. The overall sex ratio was 1.0510. Using linear and nonlinear logistic regression models with dummy variables coding for appropriately grouped municipalities, operation time periods, and corresponding spatiotemporal interactions, we consider the association between annual municipality-level birth sex ratios and minimum distances of municipalities from nuclear facilities. Within 35 km from 28 nuclear sites in France, the sex ratio is increased relative to the rest of France with a sex odds ratio (SOR) of 1.0028, (95% CI: 1.0007, 1.0049). The detected association between municipalities’ minimum distances from nuclear facilities and the sex ratio in France corroborates our findings for Germany and Switzerland. © 2016 Elsevier Inc. All rights reserved.
1. Introduction 1.1. Abundance of nuclear facilities Nuclear energy constituted 11% of global electricity production in 2011. Three countries obtain more than half their electricity from nuclear plants (France leads at 78%, followed by Slovakia and Belgium at 54% each), and ten additional countries, all but one in Europe, draw at least 25% from this source [1]. In France, 78% of the country’s electricity is supplied by the 58 currently active nuclear reactors. France is also the largest exporter of nuclear electricity in the European Union and is second only to the United States in terms of total nuclear power production contributing 16 percent to the world’s nuclear-derived electricity. While the environmental and human health risks posed by nuclear power plant accidents are well documented, modeling results by Lelieveld et al. [2] indicate that the occurrence of INES 7 major accidents and the risks of global radioactive contamination have been underestimated. Hence, human exposure risks exist around reactors in densely populated regions, notably in Central Europe and South Asia, where
∗ Corresponding author. E-mail addresses: [email protected] (H. Scherb), [email protected] (R. Kusmierz), [email protected] (K. Voigt). http://dx.doi.org/10.1016/j.reprotox.2016.02.008 0890-6238/© 2016 Elsevier Inc. All rights reserved.
a major reactor accident could subject around 30 million people to radioactive contamination. The recent decision by Germany to phase out its nuclear reactors will reduce the national risk, though a risk by reactors in neighboring countries remains. Furthermore, many nuclear facilities are 30–50 years old, contributing to the potential for catastrophic failure. 1.2. Health risks The possible health risks to populations living near nuclear facilities have prompted studies into the incidence of childhood cancer. For example, a meta-analysis of standardized incidence and mortality rates of childhood leukemia in proximity to nuclear facilities indicated elevated disease rates in the majority of the studies considered, although many findings were not statistically significant [3]. Case-control studies on juvenile cancer and leukemia were performed in Germany [4,5], Switzerland [6], Great Britain [7], and France [8]. Although these studies generally provided limited evidence (possible confounding, restricted statistical power), they nonetheless indicate an increased general human health risk in the vicinity of nuclear facilities. 1.3. Determinants of the sex ratio According to Neel and Schull, the sex ratio, or technically the sex odds, is unique among the genetic indicators [9]. Its uniqueness
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
arises from the fact that maternal exposure is expected to produce sex odds different from sex odds after paternal exposure. This expectation is attributed to the hypothesis that if an Xlinked recessive lethal gene is induced in a mother’s germ cell line by ionizing radiation, it would have no effect on a heterozygous daughter, but would be lethal to a hemizygous male zygote. Xlinked dominant lethal mutations in mothers would be equally lethal to both genders. X-linked dominant mutations induced in fathers would only suppress female offspring. Recessive X-linked lethal mutations in fathers would not influence the gender ratio as sons do not receive the paternal X chromosome and daughters carry (and are protected by) a second X chromosome from their mother [10]. In situations where mothers and fathers are nearly equally exposed on average (e.g. Chernobyl), it seems unlikely or unrealistic that the opposite maternal and paternal effects would precisely cancel each other out. That the paternal effect exceeds the maternal effect seems to be evident from increased overall sex ratios after large scale radiological incidents like the atomic bomb tests, the Windscale fire, and the Chernobyl accident [11–15]. Therefore, the odds of male to female offspring at birth may be a simple and non-invasive way to study and monitor the reproductive status or reproductive health of a population. Scholte and Sobels suggest that the observation of changes in the sex odds in the offspring of irradiated parents may be one of the few methods available for studying the genetic effects of ionizing radiation in humans [16]. Briefly, the survival probability of the female zygote is impacted by a number of lethal factors of varying degree of dominance located on the X chromosome. These factors may be impacted by radiation to either the mother or the father resulting in an impaired X chromosome, thus reducing the viability of female zygotes and changing the sex odds. In accordance with theoretical predictions, Cox found a reduced offspring sex ratio in irradiated women. James, on the other hand, states, “ionizing radiation is the only reproductive hazard, which causes (irradiated) men to sire an excess of sons” [17–19]. Ionizing radiation may differ from other causes of sex ratio variation at birth, because the effect is mediated by direct genetic intervention. It has been hypothesized that other (known) causes of variation of the sex ratio at birth may change parental hormone levels around the time of conception, which affects the probability that fertilization will be by a male sperm, or that altered hormone levels (e.g. by endocrine disrupting chemicals) may be detrimental to the survival of the male embryo [20–23]. It has also been suggested that high androgen levels in fathers (hepatitis B carriers, prostatic cancer patients) may entail male biased offspring [24]. Catalano et al. established that exposure of pregnant women to many forms of stress increases the probability of miscarriage—especially of frail male fetuses [25]. In addition to lethal factors on the X chromosome, Scholte and Sobels allude to nondisjunction resulting in X0 genotypes, which are non-viable in humans and, thus, may also distort the birth sex odds [16]. Down syndrome is a well-known consequence of meiotic nondisjunction in man, and increased Down syndrome prevalence at birth serves to indicate increased nondisjunction across Europe after Chernobyl [26]. More generally, there seems to be an association between maternal radiation history and chromosomally abnormal fetuses, e.g., Klinefelter syndrome (47, XXY) [27]. From the sex determining mechanism in man, involving the X and the Y chromosomes (XX = female, XY = male), it is obvious that the birth sex ratio must necessarily depend on three factors: • ratio of X- and Y-bearing sperm • selection of sperm within the female reproductive tract • differential implantation and survival rates of embryos
105
Investigations in the ratio of X- and Y-bearing sperm did not show significant differences in the X/Y sperm ratio, even in men with three or more same sex children [28]. Recently, Orzack et al. presented data that the sex ratio at birth is due to a higher prenatal female mortality, perhaps due to dysfunction of the paternal X [29]. If the sex ratio near or at conception is 1:1, as the extensive data by Orzack et al. indicate, but the birth sex ratio is male-biased 105:100, then female fetuses are the frailer sex (karyotype) during embryonal and fetal life, possibly mediated through their fathers’ enhanced genetic vulnerability [30]. Findings after Hiroshima and Nagasaki indicate that the sex ratio increase in the offspring of irradiated fathers per unit dose is approximately twice the sex ratio decrease in the offspring of irradiated mothers [10]. In humans, the sex odds at birth is relatively constant at the secular population level, with approximately 105 boys born for every 100 girls [31]. However, considerable variability may be observed under a variety of specific circumstances, including selective abortion in some societies. Many determinants of the sex odds, e.g., race or season as well as methodology to study those determinants have been discussed in the literature [32]. Steiner [33] points out that proposed determinants often showed associations in small samples but could not be replicated in larger populations. This, of course, may be due to insufficient statistical power due to small effects and/or small study-populations. Anthropogenic chemicals and ionizing radiation are determinants of the human secondary sex odds at birth. The effect of mutagenic chemicals or ionizing radiation on sex odds has been well established in animal experiments [34–36]. Stevenson and Bobrow [37] provide a detailed account of methodological issues relevant for the assessment of determinants of the sex odds in humans with special emphasis on the impact of male fetal mortality dynamics on the sex odds. Terrell et al. [38] reviewed approximately 100 publications on possible environmental and occupational determinants of the sex ratio. They concluded that, “limitations in study design and methodological issues make it difficult to draw firm conclusions from the existing sex ratio literature”. This highlights the general difficulties in obtaining firm knowledge about the sex odds determinants in humans. We previously studied the birth sex ratio near nuclear facilities in Germany and Switzerland, and we found evidence for increased gender proportions at birth within distances of up to 40 km from nuclear installations [13,39]. Since many anthropogenic chemicals are also mutagenic, it was natural to employ our spatial temporal methodology [40] to study the possible influence of chemical accidents on the sex odds in the vicinity of chemical plants. We specifically looked at the birth sex odds near Hoechst-Griesheim, the site of an accident in 1993 that spread tons of nitroarenes into the nearby environment [41]. We detected a decrease in the sex odds after the chemical accident [42]. Sociological influences such as stress have also been implicated in sex determination, for example following the earthquakes in Chile [43] and Italy [44]. In accordance with the Trivers-Willard hypothesis [45], these studies suggest a decrease in the human sex odds at birth under adverse living conditions. Lastly, while environmental influences play a significant role, the most dramatic determinant of human sex odds at birth seems to be man-made, namely sex selective abortion. Biased sex ratios pose a problem to societies for example in China and in India [46,47]. 1.4. Objectives Motivated by positive findings of radiation induced genetic effects after the atomic bombing of Japan [10,48], after Windscale/Sellafield [11,49], after the atmospheric nuclear weapon tests [12,13], after Chernobyl [26,50–52], and last but not least in the vicinity of nuclear facilities [39], we decided to study the
106
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
Fig. 1. Position of nuclear facilities in France, see Table 1. Nuclear power plants (grey); nuclear mining, research, storage, and disposal facilities (black).
possible association of the births sex odds with minimum distance from nuclear installations in France. France constitutes a special case in that it has comparatively many nuclear power sites, and thus provides a possibility to repeat the analyses we previously conducted for Germany and Switzerland and potentially confirm our pilot findings [13,39]. Our sex odds studies in Germany and Switzerland comprise all 23 German and Swiss nuclear power plants and 5 other pertinent nuclear facilities. In order to closely approach the methodology of our previous study and to achieve a similar statistical power, we selected all 23 French nuclear power plants complemented by 5 pertinent French nuclear mining, research, storage, and disposal facilities. We thus study 28 major French nuclear facilities in total, analogous to the 28 nuclear facilities previously investigated in Germany and Switzerland. The resulting list of nuclear facilities is presented in Table 1, and their geographical positions in Fig. 1. In Germany and Switzerland, we considered 4,906,027 births in the combined 35 km surroundings [39]. In France the corresponding number of births is 4,219,836, hence a rather similar number. Therefore, our present French study provides a similar statistical precision as our nuclear facility pilot study in Germany and Switzerland.
2. Data and statistical methods 2.1. Births by gender Complete annual French mainland gender specific birth data at the municipality level from 1968 till 2011 were produced by the INSEE “Institut national de la statistique et des études
Table 1 French nuclear power plants and major nuclear facilities. No.
Nuclear power plants
Location
Exposure since
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
BELLEVILLE BLAYAIS BUGEY CATTENOM CHINON CHOOZ CIVAUX CRUAS DAMPIERRE EL4 FESSENHEIM FLAMANVILLE G2/PHENIX GOLFECH GRAVELINES NOGENT PALUEL PENLY RAPSODIE ST. ALBAN ST. LAURENT S/PHENIX TRICASTIN
LENE BRAUD ST. LOUIS ST. VULBAS CATTENOM AVOINE CHARLEVILLE CIVAUX CRUAS DAMPIERRE-EN-BURLY BRENNILIS FESSENHEIM FLAMANVILLE MARCOULE AGEN GRAVELINES NOGENT-SUR-SEINE PALUEL PENLY CADARACHE SAINT-MAURICE-L’EXIL ST. LAURENT DES EAUX CREYS-MALVILLE PIERRELATTE
1986 1980 1971 1985 1962 1966 1996 1982 1979 1966 1976 1984 1958 1989 1979 1986 1983 1989 1966 1984 1968 1985 1979
GRENOBLE CAEN LUXEUIL KRUTH SOULAINES-DHUYS/CSA
1971 1983 1966 1953 1992
Nuclear facilites RESEARCH CENTER 24 RESEARCH CENTER 25 26 STORAGE SITE URANIUM MINING 27 WASTE DISPOSAL 28
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
107
Fig. 2. Jump function (35 km, left) and Rayleigh function (right) for the live births sex odds (male:female) depending on distance from nuclear facilities (NF) in France. The dashed Rayleigh function is the baseline adjusted Rayleigh function in Fig. 4 by Scherb and Voigt [13]; SORbaseline 1.0035 for Germany + Switzerland vs. France.
économiques” (http://www.insee.fr/fr/) and provided by the Centre Maurice Halbwachs (CMH), Partenaire du Reseau Quetelet in Paris, France (http://www.cmh.ens.fr/). The SAS data sets obtained from CHM account for 36,565 different municipalities giving rise to 1,412,109 individual municipality-years with 33,114,626 total, 16,968,701 male, and 16,145,925 female births. The overall sex odds in France from 1968 to 2011 was therefore 1.0510, (95% CI: 1.0502, 1.0517).
nuclear facilities in the German and Swiss studies. For France, we used Lambert-93, a projected orthogonal CRS that is suitable for use in France and Corsica. Lambert-93 is a CRS for large and medium scale topographic mapping and engineering survey, defined using information from IGN—Paris, “Institut national de l’information géographique et forestière” (http://www.ign.fr/). Lambert-93 is well suited for our task of determining the minimum distances of municipalities from nuclear facilities in France.
2.2. Nuclear facilities
2.4. Statistical analyses
The characteristics and geographical positions of the investigated French nuclear facilities including active and inactive nuclear power plants were obtained from the comprehensive documentation by ANDRA, the French “Agence Nationale pour la Gestion des Déchets Radioactifs—National Radioactive Waste Management Agency” (http://www.andra.fr/, http://www.andra.fr/ inventaire2012/#/documents/). Additional information on nuclear power plants, e.g., the year of the “first grid connection”, were obtained from the IAEA “International Atomic Energy Agency” information system (http://www.iaea.org/pris/CountryStatistics/ CountryDetails.aspx?current=FR). In Table 1, we list the 23 French nuclear power plants and the 5 nuclear facilities chosen, their locations, and the first year of possible exposure and effect. Due to the measurement and testing periods necessary before a nuclear power plant can enter commercial use, we set the first year of possible exposure to be one year before the year of the official “first grid connection”. As a rule, each reactor site contains between two and four reactor blocks that became operational over several decades starting in the 1960’s. Fig. 1 presents the geographical positions of the 23 nuclear power plants and the remaining 5 nuclear facilities in France.
We applied linear and nonlinear logistic regression to investigate spatiotemporal trends in the occurrence of boys among all live births in the vicinity of French nuclear facilities, and any potential change in the trends following the activation of those nuclear facilities [53]. We denote the number of male births by m, female births by f, and the fraction of male births as pm = m/(m + f). Two important parameters in this context are the sex odds (SO = pm /(1 − pm ) = m/f), and the sex odds ratio (SOR = SO1 /SO0 = (m1 /f1 )/(m0 /f0 )), which is the ratio of the sex odds between two populations SO1 and SO0, e.g., in exposed versus non-exposed populations. We used dummy coding for single points in time and for time periods as well. For example, the dummy variable for the time window from T onward is defined as dT (t) = 0 for t < T and dT (t) = 1 for t ≥ T. The simple and parsimonious logistic model for a trend and a jump in T has the following form:
2.3. Geo-coding We used the orthogonal Gauss–Krüger coordinate reference system (CRS) for the geographical locations of municipalities and
mt ∼ Binomial((m + f)t ,t ) log odds(t ) = intercept + ˛ × t + ˇ × dT (t) where t denotes the year; mt the number of boys born at time t; (m + f)t the number of live births at time t; and t the probability of a male birth at time t. Completely analogous functions may be defined for distance in place of time. Nonlinear logistic regression for fitting a Rayleigh function [13,39] was carried out by applying nonlinear regression
108
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
Fig. 3. Time trend and jump function for the sex odds within 35 km from the nuclear waste disposal site “Centre de stockage de l’Abue”, jump in the year 2000, jump SOR 1.0717, (95% CI: 1.0101, 1.1370), p-value 0.0218.
to the natural logarithm (log) of the sex odds inversely weighted by the variance of the log sex odds. The data in this study were processed with Microsoft Excel 2010. For statistical analyses, we used R 3.1.0, MATHEMATICA 9, and SAS 9.3 (SAS Institute Inc: SAS/STAT User’s Guide, Version 9.3, Cary NC: SAS Institute Inc., 2012). 3. Results 3.1. Global analyses Fig. 2 (left) presents the live births sex odds (1968 to 2011 combined) in mainland France by the minimum distance of municipalities categorized in 1 km wide distance rings around all 28 pertinent nuclear facilities depicted in Fig. 1. The sex odds ratio SOR between births within and outside of the 35 km distance is 1.0028, (95% CI: 1.0007, 1.0049), Wald Chi-square p-value 0.0079, F-Test pvalue 0.0096, see Fig. 2 (left) bold solid line. We obtain even more precise results when using an impartial Rayleigh function (F-test p-value 0.0018) with 3 parameters instead of only 2 parameters for the simple jump function. The Rayleigh function has the additional advantage of not requiring an arbitrary predefined distance such as 35 km. The estimated baseline sex ratio is 1.0506, (95% CI: 1.0499, 1.0514). The estimated sex odds peaks at 9.1 km, (95% CI: 5.8, 12.4) with a SORpeak 1.0084, (95% CI: 1.0036, 1.0132), see Fig. 2 (right). The jump and Rayleigh function analyses are based on all 33,114,626 births in France from 1968 to 2011. These findings essentially mean that within 35 km from the investigated 28 nuclear facilities in France combined, the sex odds is 1.0535 whereas outside of these combined areas, i.e., in the rest of France, the sex odds is 1.0506. If this increase in the sex odds by the factor 1.0028 (sex odds ratio) in the vicinity of nuclear facilities were exclusively due to a deficit in girls in the denominator, it would correspond theoretically to 5730, (95% CI: 1499, 9982) missing girls in the combined 35 km vicinities of the 28 nuclear facilities. We emphasize that this is probably a conservative quantification of the “real effect” as there are more nuclear facilities in France not yet investigated, and the effect probably ranges farther than 35 km. The latter can be anticipated from inspection of Fig. 2 (left): the 50 km range is also significant (p-value 0.0053). Moreover, we must assume considerable non-differential exposure misclassification biasing our results towards null. Due to the presence of different ionizing materials including neutrons, differing exposure pathways, varying susceptibility of affected people, and relevant nuclear facilities not presently analyzed, it is possible or
even likely that the effect identified in this study (Fig. 2) only partially reflects the presumable “real impact” of nuclear facilities on the secondary sex ratio in the French population. 3.2. France versus Germany and Switzerland For comparison, we included the original Rayleigh curve of our German and Swiss analysis (dashed line in Fig. 2, right) adjusted to account for the different baseline sex odds levels of Germany and Switzerland compared to France [13]. Therefore, the dashed Rayleigh function in Fig. 2 is the baseline adjusted Rayleigh function in Fig. 4 by Scherb and Voigt [13]. The baseline sex odds ratio of Germany and Switzerland compared to France is 1.0035. We previously derived the following parameters from the German and Swiss data: the jump model for distances below 35 km yields a sex odds baseline level of 1.0543, (95% CI: 1.0532, 1.0553) and a sex odds ratio for the jump at 35-km distance of 1.0036, (95% CI: 1.0015, 1.0056), p-value 0.0003. Using an impartial Rayleigh function (pvalue 0.0014), the estimated sex odds peaks at 14.3 km, (95% CI 9.1, 19.5) with a SOR peak = 1.0052, (95% CI 1.0022, 1.0082), p-value 0.0014. Therefore, the French and the German/Swiss 35-km-jump and Rayleigh curve parameters coincide reasonably well within the statistical random variability of the data. 3.3. Central nuclear disposal facility We next took a closer look at French nuclear storage sites (Table 1 and Fig. 1). Here, as in the case of the German interim highly active nuclear waste (HAW) disposal facility Gorleben [54], we found an increase of the human sex odds at birth within 35 km from the “Centre de l’Aube Nuclear Disposal Facility (CSA)” from the year 2000 onward (jump SOR 1.1467, (95% CI: 1.0615, 1.2388), p-value 0.0218. This analysis is based on 75,744 births from 1968 to 2011 in that region (Fig. 3). In order to further investigate the potentially causal relationship between the observed increase in the sex odds around the CSA from 2000 onward, we also investigated the distance dependency of the sex odds ratio (after versus before the year 2000) in 5 km distance rings. In fact, the sex odds ratio peaks within 10–30 km from the site. This effect around the CSA is similar, in shape an in magnitude, to the effect of our previous “Gorleben” analysis [39,54]. Therefore, the question arises whether neutron emitting HAW has also been stored at the CSA by the end of the 1990’s.
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
109
Fig. 4. Spatial trend of the sex odds ratio (SOR: from 2000 vs. before 2000) of aggregated gender specific live births data in 5 km distance rings. The estimated sex odds peaks at 11.6 km, (95% CI: 6.6, 16.6) with an SORpeak 1.1416, (95%CI: 1.0255, 1.2708), F-test p-value 0.0323.
3.4. Luxeuil air base Luxeuil (French: base aérienne Luxeuil) is a medium sized air base located near Luxeuil-les-Bains in the Franche-Comté region in the Northeast of France not far from the German border, see Fig. 1. From May 1966 onward, the air base hosted an operational unit for specialized ammunition storage responsible for the nuclear strike alert facilities on the base. According to the archive “La France nucléaire: matières et sites”, materials processed and stored at Luxeuil include the radioactive elements uranium, tritium, plutonium, and deuterium, see http://web.archive.org/web/20120114121607/http:// www.francenuc.org/fr sites/franche lux f.htm. For this presumed exposure situation, and since the 40 km area around Luxeuil is more densely populated, representing as much as 160,240 births from 1968 to 2011 (compared to 69,534 births around the “Centre de l’Aube” in that same period), we took a closer look at the possible association of the spatial sex odds distribution with distance from Luxeuil. Fig. 5 discloses a distinct and significant increase of the sex odds within 40 km from the nuclear air base. The optimum fit function, in this case, seems to be a (less parsimonious) shifted Gaussian that, contrary to the Rayleigh function, allows for a non-zero sex odds excess at distance zero [53]. However, for proper effect quantification we need a parsimonious approach, as for example a simple jump function at distance 40 km. The jump at 40 km (bold broken line in Fig. 5) yields SORjump 1.0203, (95%CI: 1.0102, 1.0305), p-value < 0.0001. For illustration purposes, we may note that the increased sex odds within 40 km around Luxeuil translates to the potential loss of 1570 girls (=2.03%) per 77,328 girls actually born in that area from 1968 to 2011, given that theoretically only girls and no boys were detrimentally affected by the presumed exposure around Luxeuil. 4. Discussion We found evidence for a significantly elevated human sex odds at birth in the vicinity of 28 nuclear facilities in France, in principle agreement with the results of previous analyses for Germany and Switzerland [13,39]. There is little doubt that mutagenic physical and chemical environmental or occupational hazards can alter the human sex ratio at birth. We observed consistently elevated sex ratios after Chernobyl across Europe, and even as far away as Cuba, contaminated foodstuffs from the former Soviet Union presumably caused increases in the sex ratio [13,15,50,55–57].
Recently, based on our findings, Grech [11,49] has shown that the sex ratio increased in parts of Scandinavia and especially in the most heavily exposed regions of Norway in the ten-year period following the October 1957 Windscale/Sellafield accident in the United Kingdom. Moreover, Grech [12,58] reports a rising birth sex ratio in most regions in temporal association with atomic weapon testing. Since childhood cancer and childhood leukemia are elevated near nuclear power plants, and cancer, birth defects, and sex odds increased after Chernobyl, it was natural to investigate the sex odds in the vicinity of nuclear reactors and more generally near nuclear facilities of all kinds. Based on 20 million births in Germany and Switzerland at the municipality level from 1969 to 2009, we found increased sex odds near nuclear power plants and other nuclear facilities [13,39]. Our present analyses on French nuclear facilities corroborate our previous findings in Germany and Switzerland. In France, we observed a small significant increase of a few per mill in the sex odds around nuclear facilities including all nuclear power plants, and a considerable increase near the waste disposal site CSA in SoulainesDhuys, which is similar in shape and magnitude to the increase around Gorleben, Germany. At the French nuclear air base Luxeuil, we see a long term (1968–2011) highly significant sex odds increase of 2% within 40 km from the site. The strength of our approach is that we analyze total national data and no random samples from data. Therefore, sampling error and sampling confounding is not an issue for our studies. A weakness is of course the highly aggregated nature of our data at an annual municipality basis ignoring alternative sex ratio determinants. However, this obvious drawback is perhaps more than outweighed by the amount of individual births data in the tens of millions, and the corresponding full registration over decades within all municipalities at a national basis. Moreover, it is unlikely that any alternative sex odds determinant (environmental, socioeconomic, ethnic, etc.), other than ionizing radiation, consistently entails spurious sex odds increases near practically all major nuclear facilities. The possibility of incomplete knowledge in the radiation sciences has been emphasized in the nuclear accident context. For instance, unknown exposure pathways or unsuspected radionuclides may pose a threat [59]. The second author of this study has estimated the amounts of thermal neutron activation products near casks containing highly radioactive wastes [60]. The activation product 41-Argon may play a more significant role than previously assumed. Another possibility is that the range and the biological
110
H. Scherb et al. / Reproductive Toxicology 60 (2016) 104–111
Fig. 5. Spatial trend of the sex odds (SO) of aggregated gender specific live births (1968 to 2011) in 5 km distance rings. The smooth fit function (bold solid line) is a shifted Gaussian [53] with p-value 0.0013. The jump at 40 km function (bold broken line) yields SORjump 1.0203, (95%CI: 1.0102, 1.0305), Wald Chi-square p-value
