Infant Mortality in Switzerland (1950 - 2022) and the Chernobyl Accident: Sex-Specific and Sex-Adjusted Time Trend Analyses

Background: In Switzerland after the Chernobyl accident in April 1986, the cumulative radiation dose up to 2005 was around 3,500 Sieverts, corresponding to 25 µSv/year per person. Stillbirths, perinatal mortality, and congenital malformations increased in a dose-dependent and sex-specific manner in numerous countries affected by Chernobyl fallout. Less attention has been paid to the gender specific infant mortality rate. The aim of this report is to study the secular sex-specific infant mortality trends in Switzerland (1950 - 2022) and possible trend changes after Chernobyl. 

Methods: Counts of annual live births and infant deaths under one year of age by gender for Switzerland from 1950 to 2022 were obtained from the Human Mortality Database. The overall infant death proportion is 1.13% (female 0.99%, male 1.27%), i.e., 69,905 total infant deaths in 6,186,134 total live births (female 29,677 vs. 3,010,130 and male 40,228 vs. 3,176,004). Time trend analyses of total, female, and male infant death proportions employing logistic regression were carried out. Possible level-shifts in the annual mortality rates and in the infant death vs. livebirth sex odds ratios from 1987 onward were estimated and tested. 

Results: In Switzerland, total, female, and male infant mortality rates abruptly increased in 1987 relative to the monotone secular downward trends as estimated from the period 1950-1986. The jump odds ratio in 1987 with 95%-confidence interval and p-value for the total (female + male) mortality was 1.175, (1.102, 1.253), p-value < 0.0001; females: 1.187, (1.095, 1.287), p-value < 0.0001; males 1.167, (1.078, 1.262), p-value = 0.0001. The relatively stable infant mortality sex odds ratio in the period 1950 to 1986 of 1.307, (1.283, 1.331) decreased continuously in the period after Chernobyl to a value of 1.134 in 2022, according to the 10-years sex*time*period interaction odds ratio of 0.960, (0.939, 0.982), p-value = 0.0003. 

Conclusion: The jumps in the infant mortality rates in Switzerland in 1987 and the changing annual infant mortality vs. live births sex odds ratios during the post-Chernobyl period indicate possible sex-differential contamination impacts and corroborate previous findings of increased sex-linked detrimental radiation induced genetic effects after Chernobyl.

The World Health Organization in 1957 emphasized “Man’s most precious trust is his genetic heritage, upon which must depend the health and orderly development of future generations” [1]. It is well known that ionizing radiation may induce cancer and a variety of detrimental genetic effects [2-10]. Since nutrition is a key driver in human health [11], detrimental reproductive effects like increases in stillbirths, perinatal mortality, infant deaths, congenital malformations, reduced birth weight, and distorted birth sex ratios [12-15] may be caused by Chernobyl fallout contaminating food and tap water. More recently, distinct genetic effects have been reported in the vicinity of a Swiss nuclear power plant after an INES-2 accident [16] and around a radiologically contaminated military training ground in Germany [17]. In these incidents, underestimated risks of far-reaching neutron radiation may play a crucial role. According to official estimates in Switzerland, the cumulative radiation dose up to 2005 due to Chernobyl was 3,500 sieverts. As the Swiss population between 1986 and 2005 counted on average 7 million, Chernobyl fallout caused 0.5 mSv in total per person, or 25 µSv/year. A conservative estimation in a recent report [18] based on updated radiation cancer risks [2, 19] concluded that the collective dose of 25 µSv/year and person may have caused 400 additional fatal cancer cases in Switzerland. From this perspective, it would be no surprise if infant mortality had increased in Switzerland after Chernobyl. Since no or little attention has been paid to the gender specific infant mortality rate yet [20, 21], the aim of this report is to study the long-term sex-specific and sex adjusted infant mortality trends in Switzerland with focus on possible sex dependent trend changes after Chernobyl.

The sex-specific infant mortality rate is defined as the number by sex of resident newborns in a defined geographic area (country, state, county, etc.) dying under one year of age, divided by the number of live births for the same sex and area, usually for a calendar year. Sex-specific counts of annual live births and infant mortality under 1 year of age for Switzerland from 1950 to 2022 were obtained from the freely accessible Human Mortality Database. Table 1 lists the live birth counts and the numbers of infant deaths in age category zero (0) by gender and year. The overall infant death proportion is approximately 1%: 69,905 infant deaths in 6,186,134 live births. The ascertainment error of infants being born in the year before the year of their death is considered negligible in the present context. Parsimonious time trend analyses employing logistic regression for total, female, and male infant death proportions were carried out. The following most parsimonious albeit well suited and fitting models (1) and (2) in SAS notation were applied to the total and the sex-specific data in Table 1, respectively.

(1)        model ID/LB =           t d1970*t d1970*t2 d1987    / scale=d ;

(2)        model ID/LB =           t d1970*t d1970*t2 d1987 sex sex*d1987*t    / scale=d ;

In models (1) and (2), time in years is denoted by t. For convenient numerical representations and interpretation of estimates and odds ratios t is measured in 10 years. Additionally, t and its powers as factors of the model variables are centered at the corresponding change-points in the years 1970 and 1987, respectively. dyear(t) is a dummy variable which is 0 for t < year href="#_ENREF_8">8,22]; (2) Increased stillbirth rates and sex ratio shifts were seen across Europe after the Chernobyl accident from 1987 onward [5,23]. Models (1) and (2) were derived by backward selection from corresponding initially full models containing all powers of t up to the third order and all 2nd and 3rd order interactions of those powers of t with the change-point dummy variables d1970 and d1987. An alternative but equivalent data analysis would be employing Poisson regression in place of logistic regression. This is especially important for determining how changes in sex ratios relate to changes in absolute numbers in the corresponding numerators and/or denominators [24]. However, in the present context focused on infant deaths proportions, logistic regression is the method of choice. The Wald-Chi2 statistic was used to test whether potential level shifts (jumps or kinks) from 1970 and 1987 were different from zero. A p-value less than 0.05 was taken to represent a statistically significant result. The code/data-analysis/output for this paper was generated using SAS software, mainly SAS-PROCs LOGISTIC and SGPLOT. Copyright © 2021 SAS Institute Inc. Cary, NC, USA. All Rights Reserved. SAS On Demand Release 3.1.0. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc.

Table 2 and Table 3 compile the estimates, standard errors, odds ratios, p-values, and according 95%-confidence limits of model (1) and model (2) applied to the total and sex-specific data in Table 1, respectively. Figure 1 displays the total counts (gray dots), the predicted model fit (red line), and the corresponding null-model (dotted line). The gap between the predicted line and the line of the null model corresponds to 1989 (1239, 2696) excess total female and male infant deaths from 1987 to 2022. Analogously, Figure 2 and Figure 3 show the results of fitting model (1) to the female and male data in Table 1, respectively. In Figure 2, the gap between the predicted and dotted lines indicates 924 (506,1304) excess female infant deaths from 1987 to 2022; and in Figure 3 this gap corresponds to1071 (542,1560) excess male infant deaths. The legends in Figures 1 to 3 contain additionally the respective excess counts with 95%-confidence limits for the first 7 years (1987-1983) after the Chernobyl accident. Those 7-year periods played a role in institutional discussions [18]. Whereas the highly significant main effect ‘sex’ in Table 3 with odds ratio 1.307 (1.283,1.331) means that in the period 1950-1986 approximately 30% more infant boys than infant girls died in the first year of their life, the highly significant interaction effect ‘sex*d1987*t’ means a significant gradual reduction of this gender gap by approximately 4% per 10 years: odds ratio 0.960 (0.939,0.982). Table 4 represents a dichotomized version of this observation: the annual infant mortality vs. live births sex odds ratios (male/female) decreased significantly by nearly 10% in the post-Chernobyl aera compared to before. Figure 4 is graphical representation of this finding.

Figure 1: Total (male + female) infant death rate in Switzerland and red solid linear logistic trend line according to model (1); the dotted line indicates the trend under the null hypothesis of no trend change from 1986 onward; MABTFE means assumed “Maximum Atomic Bomb Test Fallout Effect” in 1970, see [22].

Table 1: Total and sex-specific infant mortality (1 year) and live births in Switzerland 1950 to 2022.

Figure 2: Female infant death rate in Switzerland and red solid linear logistic trend line according to model (1); the dotted line indicates the trend under the null hypothesis of no trend change from 1986 onward; MABTFE means assumed “Maximum Atomic Bomb Test Fallout Effect” in 1970, see [22].

Table 2: Parameters, estimates, and pertinent statistics of model (1) for the annual total counts of Table 1.

Figure 3: Male infant death rate in Switzerland and red solid linear logistic trend line according to model (1); the dotted line indicates the trend under the null hypothesis of no trend change from 1986 onward; MABTFE means assumed “Maximum Atomic Bomb Test Fallout Effect” in 1970, see [22].

Table 3: Parameters, estimates, and pertinent statistics of the sex-adjusted model (2) for the annual female and male counts of Table 1.

Figure 4: Before vs. after Chernobyl comparison of the vital status sex odds ratio VSSOR: infant death (ID) sex odds divided by sex odds of the live births (LB) surviving the first year of live, i.e., LB-ID, the difference is significant with a period vital status sex odds ratio ratio (PVSSORR) of 1.077 (1.04, 1.12), p-value = 0.0001, see Table 4.

Table 4: Births in Switzerland 1950-2022 by sex, vital status in the first year of life, and period; sex*vital status*period 2x2x2-table; pertinent statistics for assessing the significance of the corresponding sex odds ratio ratio (SORR) sex*vital status*period interaction [31], see Figure 4.

In Switzerland, infant mortality rates are subject to significant level shifts after the Chernobyl accident in April 1986. In addition, the relatively high infant mortality vs. live births sex odds ratios (male/female) of 1.3 decreased significantly from 1987 onward. So, the question arises, whether infant death level-shifts and divergent trends in sex ratios of infant deaths versus sex ratios of live births in Switzerland after Chernobyl are sentinel indicators of radiation-induced distortions of the human genome [1,8,12]. The identified sex-differential radiation effect on infant deaths in Switzerland could be related to a finding on congenital malformations in Bavaria/Germany [15,25]. As with malformations in Germany, this genetic phenomenon seen in Switzerland may be interpreted as follows: Uncontaminated, girls present a lower level of postnatal fatal risk than boys since the genetically more vulnerable girls had already been vanishing more likely during pregnancy [8]. This ‘vulnerable females effect’ might also explain the ‘natural’ secondary sex ratio of 1.05: The primary human sex ratio seems to be 1.0 [26, 27], and the vulnerable female embryonal and fetal life entails a deficit of girls at birth. As during pregnancy, after birth again, ‘contaminated’ girls prove to be more vulnerable since the female infant mortality increased by nearly 10% relative to the male infant mortality in the post-Chernobyl period in Switzerland. These findings corroborate previous observations demonstrating elevated genetic sex-linked detriment in humans under escalated radiological conditions [13, 16, 17, 21, 24, 28]. Finally, it is necessary to emphasize, that the genetic effects identified in Switzerland from 1970 and/or 1986 onward must not necessarily be due to atomic bomb fallout or Chernobyl alone. Considerable parts of the Swiss population live within 35 km around major nuclear facilities from which persistent radiological effluents may induce cumulating detrimental albeit subclinical genetic health effects predominantly affecting potential fathers and their vulnerable female offspring[17,24,29,30]. 

The hypothesis that minute ionizing radiation exposure entails disproportional fewer female births and somewhat more previously damaged female offspring by compromising the emergence of viable babies and infants in a gender-biased manner should be investigated more thoroughly. Disproportionately lesser female and more congenitally damaged female offspring and infants would manifest as increased birth sex ratios and decreased infant death sex ratios, which is exactly what can be observed at the country-level in parts of Europe after Chernobyl.

95%-CI or (.,.): 95%-confidence interval; DF: degrees of freedom; ID: infant death, deceased at age less than 1 year; LB: live birth; Ln: natural logarithm; O: odds = 1/(1-p) for a Binomial probability p; OR: odds ratio, i.e., the ratio of two odds; ORR: odds ratio ratio, i.e., the ratio of two odds ratios; PO: period odds; POR: period odds ratio; SAS: Statistical Analysis System, software produced by SAS Institute Inc; SO: sex odds; SOR: sex odds ratio; VS: vital status   

Ethical Approval and Consent to participate

Not applicable. Ethics approval and consent to participate are not required and not necessary, since only publicly available data and previously published information is being used. 

Consent for publication

Not applicable. Only anonymous data is being used. 

Availability of supporting data 

The employed data has exclusively been published previously and/or it is contained in the Tables and in the Figures included in this paper.

Competing interests 

The author declares that he has no conflicts of interest.

Funding 
The author declares that he has no funding for this study.

Authors' contributions

Not applicable.

Acknowledgements
I am most grateful to the reviewers for detailed suggestions improving the initial draft.