A recent study published in Nature Communications reported a statistical association between living near operating nuclear power plants and higher cancer mortality rates in the US. These findings suggest a potential epidemiologic signal but leave key issues unresolved. The study did not directly assess individual radiation exposure, may not have fully accounted for confounding factors, and did not establish a causal link.

Therefore, these results should be interpreted with caution. The analysis identified association, not causation. Further research is needed to clarify the exposure pathways, better control for confounding factors, and assess the clinical and public health relevance of these findings.

Ongoing Debate

International attention has once again shifted to energy, including nuclear power generation. Past events that have shaped public perception and policy continue to influence how the issue is viewed. Concerns about safety and health effects continue to reappear in the public debate.

The association between cancer mortality and proximity to nuclear power plants is still unclear. Earlier studies have reported inconsistent findings, and our analysis adds to this ongoing discussion.

Researchers have reported higher cancer mortality rates in counties located closer to nuclear plants. The association is stronger in older adults, particularly in men aged 65-74 years and women aged 55-64 years. In some groups, the estimated relative risk was 1.19-1.20 when comparing areas with greater cumulative proximity to nuclear power plants with areas farther away from them.

The researchers estimated the mortality burden attributable to this proximity. These estimates should be interpreted with caution, as they assume a causal relationship that the study design cannot establish.

Although the analysis did not show a causal link mediated by radiologic exposure, its large scale and wide geographic and temporal scope mean that the findings should not be dismissed as anecdotal.

Study Design

The study followed a longitudinal ecological design and analyzed cancer mortality across US counties located within 200 km of operating nuclear power plants. In the absence of individual dosimetry or direct environmental measurements, researchers have used geographic proximity metrics as proxies for exposure. This metric was defined as the inverse distance to operating plants within 200 km of each county centroid or the sum of the inverse distances when more than one plant was present.

The proximity metric was averaged over extended periods to represent sustained proximity over time and to approximate cumulative exposure and potential latency. The analysis was stratified by age and sex and covered data from 2000 to 2018, while adjusting for sociodemographic, behavioral, environmental, and healthcare access variables.

Not all cancers have the same radiobiological plausibility or latency. The study did not analyze specific tumor types but assessed the aggregate mortality of all malignant cancers. Individual dosimetry and direct environmental measurements were not considered. Exposure was assigned using the county centroid, with no information on the individual’s residential history.

While highlighting certain limitations of the study’s conclusions with Univadis Spain, part of the Medscape Professional Network, Pablo Fernández-Navarro, MD, scientific researcher at the Cancer and Environmental Epidemiology Unit of the National Centre of Epidemiology at the Carlos III Health Institute in Madrid, Spain, and a member of the Consortium for Biomedical Research in Epidemiology and Public Health, said, “Among its strengths are its national coverage, long timeframe, and large sample size. Furthermore, the adjustment included potentially relevant variables such as income, poverty, racial composition, population density, smoking, BMI, temperature, humidity, and proximity to hospital care.”

“The correct interpretation of the study is that it shows a spatial association at the ecological level, not a demonstration of causality. The authors acknowledged that its design is observational and ecological, that exposure is estimated through geographic proximity rather than individual dosimetry, and that causality cannot be established. Furthermore, exposure is assigned according to the county centroid, and there are no residential histories or individual radiation measurements.”

He continued, “The finding should be interpreted as an epidemiological signal that generates hypotheses, useful for supporting better-designed subsequent studies, but not as proof that living near a power plant causes excess cancer mortality. In this type of study, one must consider ecological fallacy, potential residual confounding, and imperfect classification of exposure, because distance is only a proxy [indirect indicator] and does not equate to absorbed dose.”

Because the data are aggregated by county, age, sex, and year, the analysis does not allow for the determination of whether individuals who died lived closest to the plant, how long they had resided there, what cumulative exposure they had, or whether social, clinical, or comorbid conditions influenced the outcomes.

However, Fernández-Navarro noted: “The study does show that counties closer to central areas had higher cancer mortality rates during the period analyzed, even after adjusting for multiple covariates, and that the association was more pronounced in some age groups, especially men aged 65-74 years and women aged 55-64 years.”

Regarding the implications of the findings, Fernández-Navarro said that “mortality does not directly measure the risk of developing cancer but rather reflects the outcome after combining incidence, survival, access to diagnosis and treatment, and comorbidity. In tumors with a poor prognosis, incidence and mortality may be closer; in tumors with high survival rates or those strongly influenced by screening and treatment, this is not necessarily the case.”

Nuclear Exposure

In response to a question on the health effects of ionizing radiation near nuclear power plants, they said, “The strongest evidence links ionizing radiation to leukemia and several solid tumors, including thyroid, breast, lung, stomach, colon, liver, and bladder cancers, primarily in high-exposure and occupational cohorts. The article cites classic examples, including survivors of Hiroshima and Nagasaki, where excess leukemia was seen first, followed by increases in several solid tumors.” However, he added, “extrapolating that the relationship to chronic low-level environmental exposure near a power plant requires caution. At low doses, uncertainty is greater, the biological response may not be linear, and risk may be influenced by age, sex, lifestyle, genetic factors, and other environmental exposures.”

In other words, “The fact that ionizing radiation is a recognized carcinogen does not automatically imply that typical environmental exposure around a normally operating power plant produces a detectable increase in cancer in the population. Such an inference requires better dose measurement, identification of exposure pathways, and analysis of specific radiosensitive tumors.”

“It is also important to note that when studying total cancer mortality rather than specific tumors, neoplasms with very different sensitivities to radiation are grouped together, which can dilute or obscure the etiological interpretation.” Fernández-Navarro noted that the article itself acknowledges this limitation, as it combines cancers with different latency periods and radio sensitivities.

Key Limitations

Fernández-Navarro noted that the study could not determine whether cancer incidence increased at the same rate as the observed mortality, whether the effects were consistent across cancer types, or whether the detected pattern reflected radiation exposure rather than other geographic factors. He added that the estimated burden may not be the full burden of the disease because the study did not report the incidence, survival, years of lost life, or other outcomes.

He also noted methodological limitations. Although the models were appropriate for grouped data, they did not explicitly account for geographic heterogeneity or spatial autocorrelation, which may be relevant to strongly territorial phenomena.

Conclusions

In the debate on nuclear power and health, positions continue to range from caution to reassurance. This new study adds data that calls for further attention. However, Fernández-Navarro emphasized that the findings should be interpreted carefully because they reflect a spatial association between proximity to nuclear facilities and cancer mortality at an aggregate level, not evidence of causality.

The statistical association found should be taken seriously as a starting point for further epidemiologic research. However, it does not establish causality or measure individual exposure; therefore, it is insufficient to conclude that living near a nuclear power plant increases the risk for cancer-related death. Simultaneously, the limitations of this study do not support the opposing conclusion that a causal relationship cannot exist.

More detailed research is needed, including improved reconstruction of individual residential histories, more granular data on emissions and potential exposure pathways, and atmospheric and hydrologic modeling. Where possible, studies should incorporate direct dosimetric estimates. In addition, analyses of the incidence of cancer and specific tumor types, particularly those with stronger radiobiological plausibility, would help clarify the observed association.

Fernández-Navarro reported having no relevant conflicts of interest.

The original study authors reported having no relevant conflicts of interest.

This story was translated from Univadis Spain.