Mortality of populations potentially exposed to ionising radiation, 1953–2010, in the closed city of Ozyorsk, Southern Urals: a descriptive study

Background The city of Ozyorsk (Southern Urals) was created as a secret city in 1945 and is a closed city until today. It housed workers of the earliest and one of the country’s largest nuclear facilities. Workers of the nuclear reactors, radiochemical or reprocessing plants were exposed to high levels of ionising radiation in the early years of operation and possibly further exposed from inhalation of plutonium aerosols. Methods The cause-of-death registry of Ozyorsk received paper copies of original death certificates of all deaths of residents of the city. Data were analysed for recent mortality rates (1998–2010) and time trends in age-standardised mortality rates between 1953 and 2010 of main groups of causes of deaths, in particular cancer. Results Comparing workers of the three main plant types with the remainder of the Ozyorsk residents, and with national figures, all-cause mortality rates were lowest among workers, with ratios compared to national figures of 0.65 (men) and 0.56 (women), and compared to the other residents of 0.77 (men) and of 0.74 (women). For cancer overall, the differences were smaller in men (ratio between workers and national figures of 0.86) and there were no differences in women (ratio of 1.00), but ratios differed by cancer type. Most cancer deaths were however least common in the workers, including leukaemia. Over the last 60 years, all-cause mortality has gradually increased among men in all three groups but was stable among women, whereas cancer death rates have slightly declined in both sexes. Conclusions Healthy worker effect, relatively better living conditions in Ozyorsk and healthier lifestyles may explain the lower mortality rates in Ozyorsk. Overall mortality time trends in Ozyorsk were similar to the entire country. No apparent radiation-related effects were seen in this population-level analysis, but the radiation-related risks can be better addressed in individual-level studies. Electronic supplementary material The online version of this article (doi:10.1186/s12940-015-0078-8) contains supplementary material, which is available to authorized users.


Background
The first Soviet production facility of nuclear weapons, Mayak Production Association (Mayak PA) started operation in the Southern Urals in 1948 [1,2]. The main plants of Mayak PA consist of several nuclear reactors, a radiochemical plant, and a reprocessing plant. Since commissioning, the workforce exposure to ionising radiation has been continuously monitored. Practices of workers' radioprotection evolved considerably over the years of operation of the facility. In the early years of operation (1948)(1949)(1950)(1951)(1952)(1953), workers were exposed to very high doses. The workforce's mean annual doses were around 200 mGy or higher until 1953 for external exposures, with gradual decreases afterwards in the 3 main plant types. In the last 20 years, the 3-plants workers' mean annual dose of external exposures did not exceed 5 mGy. Some individuals accumulated doses of external exposures of up to 10 Gy during the prolonged period. In addition, some workers were at risk of inhaling plutonium aerosols, with the highest rates of plutonium intake in years 1949-1958 when individual devices for protection of respiratory organs were not available. About a quarter of the workforce were women.
All workers employed at Mayak PA lived in a closed city 5 to 10 km away from the facility, called first Chelyabinsk-65, then, from 1966 onwards, Chelyabinsk-40 and Ozyorsk since 1994. The city was created in 1945 for the purpose of housing current and former Mayak employees and their families; in 2013, the still closed city had about 81 000 inhabitants. This city was formerly secret, and population movements in and out of town were recorded and restricted. During the 1950's, 34 % of those working there at the time left the city, partly due to migration of some specialists to newly opened similar plants in the area, partly due to removal of some of the workforce in relation to the severe working conditions at Mayak PA at the time, and partly due to less restricted conditions for departure from the city. In contrast, since the 1970's, the worker's population was stable (0-1 % migration). Social and economic conditions of residence in Ozyorsk (salary, apartment provided etc.) were much better than in other places of residence.
The cause-of-death register of Ozyorsk was established at the Southern Urals Biophysics Institute (SUBI) in 1988 [3]. Based on copies of the original paper death certificates, it included deaths from all causes in the city since 1948. The completeness of data in the register is virtually 100 %. Additionally, quality assurance checks of the cause-of-death information have been performed for Mayak workers in the framework of previous epidemiological studies [4,5] in which it was found that 80 %, 25 %, and 50 % of death certificates were confirmed by autopsy before 1990, during 1990 -2000, and after 2000, respectively. Malignant neoplasms were found to be histologically verified for 80 % of all cancer deaths.
The objective of this study was to describe the mortality of the population of Ozyorsk and cancer mortality in particular. We compared the causes of deaths of workers of the 3 main plant types (nuclear reactors, radiochemical plant, reprocessing plant) with those of the remainder of the Ozyorsk population and with the national figures of the Russian Federation in the most recent time period (1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010). We also described long-term time trends of the mortality rates in these populations .

Data
Numbers of deaths in people aged 18 to 74 years between 1953 and 2010 were obtained from the Ozyorsk cause-of-death register [3]. They were subdivided into 2 population groups covering the entire closed city: the people ever employed at one of the 3 main Mayak plant types (reactors, radiochemical and reprocessing plants), and the remainder of the Ozyorsk population. The latter also included the other Mayak PA workers not working at any of the 3 plant types. The two groups will be called the "3-plants workers" and the "other residents" for simplicity. For the 3 plants workers, size of the population at risk by 5-years age groups for each 5-year calendar time periods was computed from the Mayak PA employment records. For the other residents, the size was computed from the lists of electors of city and state authority elections containing all the city citizens aged 18 years or older for the period before 1958, and later, from the census information in years 1959, 1970, 1979, 1989, 2002 and 2010 and the local statistics bureau for population information, from which the 3-plant workers population was subtracted. Only Ozyorsk residents contributed person-years at risk until death, migration or 2010, whichever came first. People working at Mayak PA contributed person-years to the 3-plants workers population from the time when they started working in one of the 3 plant types. For comparison, deaths rates in the Soviet Union until 1991 and in the Russian Federation afterwards were obtained from the WHO database [6] for the longest possible period of time.

Statistical analysis of observed mortality rates
Age standardised mortality rates (ASR) of the various groupings of causes of death per 100,000 person-years were calculated separately for men and women. The European standard population, truncated to 18-74 years, was used for age standardisation. Ninety-five percent confidence intervals were computed using the gamma approximation [7]. For the most recent time period, 1998-2010, the standardised rate ratios (SRR) were the ratios of ASR in 3-plants workers compared to other residents and to the national figures, and in the other residents compared to the national figures. Confidence intervals of the SRR did not account for the variability of the national figures due to their large population size. When numbers were smaller than 250 person-years in a specific age and time period group, this age group was merged with the neighbouring age group to enlarge it. Truncated standard population weights were then adapted accordingly.
All causes ASR was lower among 3-plants workers than among other residents, with a SRR of 0.77 (95 % CI 0.73, 0.81) in men and of 0.74 (95 % CI 0.66, 0.83) in women. The leading cause-of-death was from circulatory system diseases, with the rates being similarly lower than for all causes mortality, namely of 0.74 (95 % CI 0.68, 0.81) in men and of 0.75 (95 % CI 0.64, 0.88) in women.
Injury and poisoning was the third most common cause-of-death among male workers, while it ranked second among other male residents, resulting in a SRR of 0.72 (95 % CI 0.63, 0.83)). Few women died of injuries and poisoning. ASR from all other groups of causes of deaths were lower among the 3-plants workers compared to the other residents.
Mortality rates in Ozyorsk were generally lower than the national figures of the Russian Federation. The all  Longer term time trends were based on 18,232 deaths in 2,184,608 person-years in the period 1973-2010. They showed similar patterns across the three populations, despite some abrupt fluctuations over time in all cause and non-cancer mortality rates (Table 3, Fig. 1, and Additional file 1: Figure S1). National figures showed the most marked changes which appeared to be attenuated in 3-plants workers (Fig. 1a, c, d). Sixteen pleural cancer deaths were observed (Additional file 1: Figure S1, K).
In women, the time trends were less strong, but not dissimilar to those in men. Increases were observed for other causes of deaths (annual increases of 3.5 % in 3plants workers and 4.6 % in other residents), digestive system diseases (2.0 % in 3-plants workers and 2.1 % in other residents), and breast cancers (2.2 % in 3-plants workers and 1.6 % in other residents) (Fig. 1e, Additional file 1: Figure S1 V,M). Genitourinary cancers increased in 3-plants workers (2.8 %) and decreased slightly in other residents (−0.6 %). The overall cancer rates were generally stable, with decreases observed in rates of stomach cancer ( −2.2 % in 3-plant workers and −2.9 % in other residents) and lung cancers (−1.1 % in 3-plants workers and −0.5 % in other residents) (Additional file 1: Figure S1 D, I). Haematological malignancy rate was twice as high among 3-plants workers (ASR =19 based on 20 observed deaths) than among other residents (ASR = 9), but numbers were too small to compare the time trends.

Discussion
Overall, between 1998 and 2010, the Ozyorsk population showed lower all cause and cause-specific mortality rates than the national figures of the Russian Federation, and 3-plants workers showed even lower mortality rates than the other residents for most causes of deaths. This overall effect is mainly related to lower rates of deaths from circulatory system diseases, and from injury and poisoning. Cancer mortality rates showed smaller differences. Exceptions to this general pattern of lower rates in the Ozyorsk population were seen for the heterogeneous group of other causes of deaths and a few individual cancer sites. Cancer sites which showed elevated rates comparing 3 plants workers and the other Ozyorsk residents were, among men, cancers of other and unspecified sites, and cancers of bone, skin and connective tissue and in women, liver, rectum and lung cancer, multiple myelomas and genitourinary cancers. All these observations were also generally true for the period 1973 to 2010.
Mortality is a result of the frequency of the underlying diseases or events, survival from disease and how well causes of death are registered. One motivation for investigating the mortality of the city of Ozyorsk is the assumed higher exposure to ionising radiation, especially of workers of the nuclear facilities, but perhaps also the other residents due to nuclear accidents in the 1950s [1]. Ionising radiation exposure is known to increase the risk of various cancers [8], and the risk of circulatory system diseases [9]. Three plants workers had a higher average occupational dose than the other residents, although no final documentation of ionising radiation exposure of  1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005  other residents was available [10]. Dose-response relationships have been observed for lung, liver and bone cancers, for most solid tumours and for cerebrovascular and ischemic heart diseases [4,5,11,12] in Mayak workers. In our analyses at the group level, the ionising radiation exposures did not translate into higher deaths rates for 3 plants workers for most cancers. This was also not the case for death from circulatory diseases. However, some of the cancer sites showing elevated mortality rates in the 3 plants workers had been previously linked to ionising radiation exposure in this cohort (lung and liver cancer in females, bone cancer in men) [11,13,14]. Lung, liver and bone were organs with highest plutonium deposition. This likely reflects the long term impact of exposures received in the past, as the induction time of ionising radiation on solid cancers might be long. The interpretation of the other elevated rates was less clear: the high rates of the heterogeneous group of other causes of death and their divergence from the national figures deserves further attention. Death from unknown or unspecified neoplasms could be related to autopsy rate together with differential criteria for recording causes of deaths between 3-plants workers and other residents [3,14]. The rate of haematological diseases was comparable between 3 plants workers and national figures. However, while the leukaemia rates were decreased, there was an excess of multiple myelomas. They have not often been addressed on their own [11]. Among 40-79 years old, more than 30 % of male and 5 % of female deaths are attributable to tobacco smoking in the Russian Federation between 1980 and 2010 [15]. Alcohol consumption, especially vodka, is a major cause of the high risk of premature death in Russian men [16]. Given the weight of these preventable risk factors, lower prevalences in Ozyorsk and among 3-plants workers especially could explain the lower rates of cardiovascular diseases deaths, of male liver cancers, of digestive system diseases (which includes liver cirrhosis) and of injury and poisoning (including fatal alcohol poisoning and car accidents). We observed that the 7-years decrease in male life expectancy between 1987 and 1994 in Russian Federation [17], and the impact of anti-alcohol campaigns (1985-1987 and since 2005) [18] on cardiovascular mortality and on injury and poisoning deaths decrease in life expectancy was less strong in Ozyorsk and even more in 3-plants workers. This could be due, in part, because 3plants workers always had a lower alcohol consumption due to their occupation, but no information was available about this for this study.
Higher living standards in Ozyorsk in the 1960's, 70's and 80's in terms of food access and lodging than elsewhere in the Russian Federation/Soviet Union, healthy worker effect whereby persons selected into the workforce are in better general health than the general population, and maintaining a better health through better health surveillance may have contributed to lower deaths rates of the 3-plants workers compared to the other residents. The quality and accessibility of the health care in an organised and affluent city, and for the 3-plants workers specifically, could play a role in the generally lower rates of death compared to Russian Federation, if survival rates were better for the local population than nationally. The very low rate of deaths from infectious diseases, and the low rate of injuries and poisoning deaths could support this explanation, but it could also in part be linked to the isolation of the Ozyorsk population. Affluence might add to the plausibility why some death rates are lower in the Ozyorsk city compared to Russian Federation. Although medical care is free and has been free in the past, cost of medications can be borne by patients, which could influence treatment adherence of chronic conditions (eg hypertension) [19], and cancer survival. Problems of drug supply, even for patients entitled to reimbursement of anti-cancerous treatments have been reported in the context of cancer care in Russian Federation [20]. The higher rates of deaths from multiple myelomas could reflect access to better diagnostic tools for this population compared to the general population, which could impact classification, or differential coding habits. Indeed, the validation of the coding of 246 deaths certificates in the Ozyorsk cause-of-death registry showed that out of the 7 deaths with discrepant coding in the main ICD group, 2 concerned haematological malignancies [3]. The difficulty of the coding of haematological malignancies should not be underestimated.
Finally, there are some uncertainties in the underlying data. Population data is more accurate in census years than between censuses. In addition, the causes of deaths of persons who migrated from the city is unknown. These are limitations especially in the early years where a large part of the workforce was taken from Mayak PA due to the severe working conditions. Deaths among these workers were therefore not  recorded in the Ozyorsk cause-of-death registry. Since the 1970's, migration is very low. Although it cannot be ruled out that some Ozyorsk people with severe disease moved elsewhere, there is no evidence that health per se is a reason for migration.

Conclusion
Overall, mortality in Ozyorsk was lower than for entire Russia/Soviet Union, suggesting more healthy life in the close city. Mortality patterns of cancers of the most recent years do not suggest a particularly strong radiation effect and only individual-level studies with individual dose estimates are informative with regard to potential radiation-related problems.
Additional file