A national analysis of the impact of proximity to nuclear power plants on lung, breast and colon cancer mortalities in the U.S., 2000–2020

Abstract

BACKGROUND:

Nuclear power plants emit low levels of ionizing radiation, an established risk factor for breast, colon, and lung cancers, yet the long-term effects of chronic environmental exposure in U.S. populations remain unclear.

OBJECTIVE:

To evaluate sex- and age-specific associations between proximity to nuclear power plants and mortality from the three most common cancers in the U.S.: breast, colon, and lung cancer.

METHODS:

We quantified county-level proximity to nuclear power plants using the sum of inverse distances from each residence county’s population-weighted center to all plants within 200 km, updated annually from 2000 to 2020. Cancer-specific mortality data (breast, colon, and lung) from the CDC were analyzed by sex and five age groups (45–54, 55–64, 65–74, 75–84, 85 +). Relative risks (RRs) were estimated using generalized estimating equations with a Poisson link. Models were adjusted for sociodemographic factors, urbanicity, region, and temporal trends.

RESULTS:

Proximity to nuclear power plants was associated with elevated mortality from breast, colon, and lung cancers. From 2000 to 2020, an estimated 39,767 female deaths (95% CI: 9312–69,381), representing 2.01% (95% CI: 0.47–3.50%), and 38,124 male deaths (95% CI: 16,106–59,600), representing 2.33% (95% CI: 0.98–3.64%), were attributable to this proximity. Lung cancer accounted for the largest burden in both sexes, followed by breast and colon cancer in females and colon cancer in males. Mortality risks declined with increasing distance, becoming negligible beyond 50 km.

SIGNIFICANCE:

This national-scale analysis provides new evidence that proximity to nuclear power plants is associated with increased mortality from major cancers in the U.S. The magnitude and consistency of the findings highlight the importance of updated risk assessments, sustained surveillance, and strengthened public health planning for communities living near nuclear facilities.

IMPACT:

• This study provides the first national assessment of sex- and age-specific mortality from breast, colon, and lung cancers in relation to proximity to U.S. nuclear power plants, revealing consistent patterns not previously demonstrated. These findings fill a major gap in environmental epidemiology and underscore the need for cohort studies, refined exposure assessments, and pathway-specific analyses to strengthen causal interpretation. As nuclear power gains momentum in national energy planning, establishing clearer evidence on potential health impacts is increasingly essential for guiding research priorities and public health preparedness.

INTRODUCTION

Nuclear power remains integral to the global energy landscape, supplying reliable electricity lower in carbon than fossil fuels in an era characterized by rising energy demands and climatic concerns. Worldwide, nuclear reactors produced roughly 2600 terawatt-hours of electricity in 2023, with over 440 reactors operational globally and additional reactors planned or under construction (World Nuclear Association, 2023). The United States, the world’s largest producer of nuclear-generated electricity, currently operates 93 commercial reactors at 54 plants across 28 states, producing approximately 30% of the world’s nuclear-generated electricity (U.S. Energy Information Administration, 2023). Despite these benefits, nuclear power generation raises persistent concerns related to radioactive emissions and associated health risks, particularly cancer.

Ionizing radiation emitted from nuclear power plants disperses into surrounding environments through atmospheric and hydro-logical pathways, potentially causing prolonged low-dose exposure in nearby populations [1]. The cancer risks from moderate-to-high radiation doses are well documented, largely based on studies of atomic bomb survivors and medically exposed populations [2–4]. However, the effects of chronic low-dose radiation, typical of environmental exposures near nuclear plants, remain debated. Since 1959, the International Commission on Radiological Protection (ICRP) has adopted the linear no-threshold (LNT) model, which assumes that cancer risk is directly proportional to radiation dose, even at low levels—implying that no amount of radiation exposure is completely without risk. This model has also been adopted by the National Council on Radiation Protection and Measurements (NCRP), reinforcing its status as the prevailing framework for radiation protection guidelines [5–8].

In 2022, breast, lung, and colon cancers accounted for approximately 30% of the global cancer burden, with breast cancer as the most frequently diagnosed among women (2.3 million cases), lung cancer the most common overall (2.5 million cases), and colon cancer a major contributor with over 1.1 million new cases [9]. Exposure to ionizing radiation has been identified as a risk factor for all three cancers, with epidemiological studies linking medical imaging, occupational exposures, and environmental sources to increased risk [10, 11]. Although radiation-related risks vary widely depending on factors such as dose level, timing, and exposure duration, current evidence generally supports a linear dose–response relationship, reinforcing concerns about potential cumulative effects of chronic low-dose radiation exposure.

While nuclear power plants are a known source of ionizing radiation, and ionizing radiation is an established risk factor for lung, breast, and colon cancer, studies investigating a direct link between nuclear power plants and these cancers have produced heterogeneous results. Several international studies have found no significant association between cancer incidence and residential proximity to nuclear plants [1, 12–14]. while others have reported increased cancer risks, particularly among populations living closer to these facilities [15–17].

In addition to these studies, we recently reported evidence of increased cancer incidence associated with residential proximity to nuclear power plants in a population-based study of Massachusetts residents, using an inverse-distance weighted exposure metric and age- and sex-stratified analyses [18]. That study demonstrated elevated incidence for multiple cancer sites, including lung, breast, and colorectal cancers, particularly among older adults. However, that analysis was limited to a single state and focused on cancer incidence rather than mortality.

At the national level, our earlier study [19] investigating the association between proximity to nuclear power plants and all-cancer mortality, we found that an estimated 115,586 cancer deaths (95% CI: 56,964–173,326) in the U.S. between 2000 and 2018 were attributable to nuclear power plants. The burden was most pronounced in older age groups, particularly among males aged 65–74, with 20,912 attributable deaths (95% CI: 12,591–29,109), representing 2.0% (1.2%, 2.7%) of total cancer mortality in that stratum. Attributable fractions generally ranged between 1–2% in older adults, with narrower confidence intervals and higher absolute counts in the elderly.

In the United States, most research has focused on individual plants or localized areas, making it difficult to assess nationwide trends [20–27]. As a result, significant knowledge gaps remain, particularly in large-scale national studies and in understanding how cancer-specific mortality risk near nuclear plants varies by sex and age group.

To address these critical gaps, our aim in this study was to conduct the first national-scale assessment of sex- and age-specific mortality from breast, colon, and lung cancers in relation to distance from nuclear power plants, using a refined, spatially resolved, population-weighted proximity metric. We analyzed cancer mortality data from 2000–2020 obtained from the Centers for Disease Control and Prevention (CDC), covering all U.S. counties within 200 km of nuclear power plants. Unlike prior geographically limited studies, this work was designed to provide cancer-specific mortality estimates associated with nuclear plant proximity. By stratifying analyses by sex and age, we sought to evaluate differential vulnerabilities across subgroups, with the ultimate goal of informing future epidemiologic research and public health surveillance.

DATA AND METHODS

Data

Nuclear power plants.

We sourced data on nuclear power plant locations and operational timelines from the U.S. Energy Information Administration (EIA) website. To ensure a comprehensive analysis, we also included facilities in Canada that are situated within 200 km of a U.S. County population center. Whenever possible, we cross-referenced information with plant-specific websites to confirm locations and operational details. A complete list of all plants, along with their operational years and reactor types, is provided in Table 1S of the supplemental materials.

Cancer mortality

Cancer mortality data for the contiguous United States (2000–2020) were obtained from the Centers for Disease Control and Prevention (CDC). We used the “detailed mortality – All Counties” files for the years 2000–2020 from the National Center for Health Statistics [28]. We extracted unsuppressed mortality records for breast (C509), colon (C189), and lung (C349) cancers, ensuring full data coverage across all U.S. counties with population centroids within 200 km of a nuclear power plant (1922 counties). Unlike the publicly available CDC WONDER database, which suppresses county-level data with fewer than 10 deaths to protect privacy, this dataset provides complete mortality records (including county-year records with less than 10 deaths), allowing for a more accurate and geographically comprehensive analysis of cancer mortality patterns in relation to nuclear plant proximity.

Our analysis examined cancer mortality for the three leading causes of cancer deaths between 2000 and 2020, focusing on breast, colon, and lung cancers across five age groups (45–54, 55–64, 65–74, 75–84, and 85 +), stratified by sex. A summary of cancer deaths by age group and sex across the 1,922 counties within 200 km of a nuclear power plant is provided in Table 1.

Table 1.

Total cancer deaths in the 1922 counties within 200 km of an operational power plant for the duration of 2000 to 2020.

Deaths
Sex	Age Group	Lung	Breast	Colon
Male	45_54	94,257	-	29,026
Male	55_64	278,785	-	60,623
Male	65_74	432,803	-	84,009
Male	75_84	385,499	-	91,217
Male	85+	128,757	-	52,196
Female	45_54	74,344	90,216	22,949
Female	55_64	198,022	135,145	43,811
Female	65_74	321,277	139,042	64,722
Female	75_84	323,392	138,318	94,397
Female	85+	138,302	105,584	90,641
	Sum	2,375,438	608,305	633,591

Covariates

We adjusted for a set of annual county-level covariates (2000–2020) to account for potential confounders in the association between proximity to nuclear power plants and cancer mortality. Socioeconomic factors included educational attainment (percentage of the population with less than a high school education), median household income, poverty level, and percentage of renters. Demographic variables accounted for racial composition (percentages of White, Asian, and Black populations) and age structure (percentage aged 65 years and older). Urbanicity was represented by public transportation use, and housing characteristics were captured by median property value. All of which were obtained from the U.S. Census for years 2000, 2010, and 2020 [29–31]. Additionally, we adjusted for temporal and geographic variation by including fixed effects for year and U.S. Census regions (New England, Middle Atlantic, East North Central, West North Central, South Atlantic, East South Central, West South Central, Mountain, and Pacific). A complete description of these covariates is provided in the Supplementary Table 2S.

Proximity assessment

To estimate nuclear proximity at the county level, we employed a distance-based approach, calculating the inverse of the distance (1/d) in kilometers for each plant-county pair, considering all nuclear plants that were operational for at least one year between 1990 and 2020 and located within 200 km of a county’s population center rather than its geographic centroid. This approach resulted in 1922 counties within the 200 km range, with 76 of the nuclear power plants listed in Table 1S included in the analysis.

For each plant-county combination, we first calculated 1/d for each year the plant was active. Following plant decommissioning, we assigned a value of 0 for 1/d from that year onward. To account for cumulative proximity and potential latency effects, we applied a 10-year moving average to each plant’s 1/d values. Finally, we aggregated these smoothed values across all contributing plants within 200 km of a county, generating an annual proximity estimate.

As this continuous, distance-weighted metric was defined as the sum of the inverse distances (1/d) from each county’s population-weighted center to all nuclear power plants within 200 km. By construction, higher values indicate greater cumulative proximity to nuclear facilities, since counties nearer to more plants contribute larger sums than those farther away. By incorporating historical proximities, this approach ensures that past proximity remains relevant, aligning with the latency periods of radiation-associated cancers and enabling a realistic assessment of long-term exposure trends.

Statistical analysis

We examined annual lung, colon, and breast cancer mortality data from 2000 to 2020 across all U.S. counties with population centers located within 200 km of a nuclear power plant to assess the relationship between nuclear power plants proximity and cancer mortality counts. To estimate this association, we applied separate Generalized Estimating Equation (GEE) Poisson regression models for each cancer type (Eq. 1), adjusting for a range of county-level covariates detailed in Supplementary Table 2S.

The dependent variable in our models was the annual number of deaths for each cancer type, stratified by sex and age group (45–54, 55–64, 65–74, 75–84, and 85 +). To account for differences in population size across counties and demographic subgroups, we incorporated a natural log population offset, ensuring that mortality rates were appropriately scaled based on age, county, sex, and year.

To account for correlation between repeated observations within the same county over time, we specified county as the clustering variable, assuming an independent correlation structure, where all observations within a county share the same correlation. Additionally, we employed robust (sandwich) standard errors to adjust for within-county correlation, over dispersion, and any potential model misspecification, ensuring valid statistical inference.

$$log\left({\lambda }_{ijk}\right)={\beta }_0+{\beta }_1\cdot C{1}_{ij}+{\beta }_2\cdot C{2}_{ij}+{\beta }_3\cdot C{3}_{ij}+{\beta }_4\cdot C{4}_{ij}+\cdots +{\beta }_p\cdot C{p}_{ij}+{\beta }_e{E}_{ij}+\mathit{\text{log}}\left(\mathit{\text{populatio}}{n}_{ijk}\right)$$ (1)

where:

• i indexes counties, j indexes years, and k indexes age-sex groups.

• λ_ijk represents the expected number of cancer deaths in county i at year j for age group k.

• C_nij denotes covariate n for county i and year j.

• E_ij represents the sum (over plants) of inverse-distance proximity metric and β_e is its estimated effect.

• β₀ is the intercept, representing the baseline log cancer mortality rate.

We stratified our analysis by cancer type, sex, and age strata (45–54, 55–64, 65–74, 75–84, and 85 +). For each combination of cancer type, sex, and age group, we ran separate GEE Poisson regression models to evaluate the association between nuclear power plants proximity and cancer mortality.

To estimate relative risks (RRs) across varying distances, we constructed a dataset of incremental distances up to 50 km. The RR at a given distance was calculated as exp (coefficient/distance), with corresponding 95% confidence intervals derived using the same approach applied to the upper and lower bounds of the coefficient estimate.

for each cancer type, county, age group, sex, and year based on the yearly proximity values, we calculated annual RRs associated with that year’s proximity as compared to absence of proximity as the reference group (outside the 200 km range). These RRs were then used to compute attributable mortality fractions (AFs) (Eq. 2), representing the proportion of cancer deaths linked to nuclear power plants proximity.

$$AF=\frac{RR-1}{RR}$$ (2)

To determine the total number of cancer deaths attributable to nuclear proximity from 2000 to 2020, we first calculated yearly attributable deaths separately for each cancer type, county, age group, sex, and year by multiplying the attributable fractions by the number of deaths in each stratum. We then summed these values across all cancer types, counties, and years to obtain the total number of attributable cancer deaths for each cancer type, age group, and sex over the study period.

Sensitivity analysis

To assess the robustness of our findings, we conducted a Leave-KOut (LKO) sensitivity analysis [32]. where we randomly removed a subset of counties and re-estimated our models to evaluate the stability of our effect estimates. Specifically, for each cancer type, sex, and age group, we randomly excluded 200 counties (approximately 10% of the 1922 counties in our dataset) per iteration and repeated this process 100 times. This approach allows us to examine the variability of our estimates and ensures that our results are not unduly influenced by a few specific counties.

To assess whether nuclear reactor type influenced cancer mortality risk, we categorized nuclear power plants into Boiling Water Reactors (BWR), Pressurized Water Reactors (PWR), and other reactor types. We then constructed a categorical variable representing the dominant reactor type associated with each county-year observation and included this variable as an additional covariate in the models. This approach allowed us to evaluate whether differences in reactor technology were associated with cancer mortality, independent of overall proximity to nuclear power plants.

As an additional negative-outcome control analysis, we fit the same models using accidental deaths as the outcome. Accidental causes of death were defined using ICD-10 codes V01–V99 (transport accidents), W00–W99 (other external causes of accidental injury), X00–X59 (other accidental injuries), and Y85–Y86 (sequelae of accidents). These causes of death are not biologically related to low-dose ionizing radiation exposure, so the absence of associations provides a falsification test for our modeling strategy.

RESULTS

Exposures

Figure 1 displays county-level nuclear exposure estimates for the year 2000, visualizing the 10-year moving average of cumulative inverse-distance proximity (in kilometers) from operational nuclear power plants. These estimates were derived by summing the inverse-distance weights of all nuclear plants located within 200 km of each county’s population center.

Fig. 1. County-level nuclear power plants proximity (10 year moving average) metric at county level in the U.S.

(Year 2000).

Cancer mortality

Figure 2 illustrates the estimated associations (regression coefficients) between nuclear proximity and cancer mortality, displaying effect estimates and 95% confidence intervals (CIs) for breast, colon, and lung cancers across age groups, stratified by sex. The top panel represents females, while the bottom panel represents males. We observed a positive association between nuclear power plants proximity and cancer mortality, with notable differences by cancer type, age group, and sex. Among females, the strongest associations were found for lung cancer in the 55–64 age group and colon cancer in older age groups (75–84 and 85 +), while breast cancer associations increased with age, peaking in the 85+ group. In males, the highest effect estimate for colon cancer was observed in the 75–84 age group, while the strongest association for lung cancer appeared in the 65–74 age group.

Fig. 2. Sex- and cancer-specific associations between nuclear power plant proximity and cancer mortality across age groups (2000–2020).

Effect estimates (proximity coefficients) and 95% confidence intervals for the associations between proximity to nuclear power plants and cancer mortality by age group and sex. The upper panel shows results for females and the lower panel shows results for males. Purple diamonds represent breast cancer, orange circles represent colon cancer, and blue triangles represent lung cancer. The dashed horizontal line indicates the null effect estimate.

The relationship between nuclear proximity and cancer mortality risk as a function of distance from nuclear power plants and stratified by cancer type, age group, and sex is illustrated in Fig. 3. The plots indicate a decreasing relative risk with increasing distance, with the highest risks observed in counties closest to nuclear plants. The decline in risk is most pronounced within the first 50 km, beyond which the effect appears to diminish and stabilize near zero.

Fig. 3.

Adjusted relative risks (Eq. 1 & 2) of lung, colon, and breast cancer mortality by one plant equivalent distance from a nuclear power plant, stratified by sex and age group (2000–2020).

In both males and females, all cancers studied exhibited a distance-dependent attenuation of risk, with higher relative risks observed closer to nuclear power plants and declining risk with increasing distance. This pattern suggests that geographic proximity to nuclear power plants plays a key role in cancer mortality, particularly within 30 km of nuclear facilities, where the impact appears to be most pronounced.

Table 2 presents the total cancer deaths, attributable fractions, and attributable mortality for breast, colon, and lung cancers, stratified by sex and age group for all of the 1922 counties within 200 km of a nuclear power plant. Across both males and females, lung cancer accounted for the highest number of total deaths, followed by colon and breast cancer.

Table 2.

Estimated adjusted nuclear power plant proximity attributable lung, colon and breast cancers mortality and fractions, stratified by age group and sex (2000–2020).

SEX	CANCER	AGE GROUP	TOTAL DEATHS	ATTRIBUTABLE MORTALITY	ATTRIBUTABLE FRACTION
Female	Breast	45_54	90,216	1043 (−116, 2176)	1.16% (−0.13%, 2.41%)
Female	Breast	55_64	135,145	959 (−720, 2603)	0.71% (−0.53%, 1.93%)
Female	Breast	65_74	139,042	2210 (672, 3719)	1.59% (0.48%, 2.67%)
Female	Breast	75_84	138,318	2975 (1346, 4570)	2.15% (0.97%, 3.3%)
Female	Breast	85+	105,584	2653 (1297, 3979)	2.51% (1.23%, 3.77%)
Female	Colon	45_54	22,949	−86 (−567, 379)	−0.37% (−2.47%, 1.65%)
Female	Colon	55_64	43,811	574 (−245, 1366)	1.31% (−0.56%, 3.12%)
Female	Colon	65_74	64,722	592 (−375, 1534)	0.91% (−0.58%, 2.37%)
Female	Colon	75_84	94,397	1480 (−55, 2973)	1.57% (−0.06%, 3.15%)
Female	Colon	85+	90,641	2296 (1026, 3536)	2.53% (1.13%, 3.9%)
Female	Lung	45_54	74,344	2221 (535, 3840)	2.99% (0.72%, 5.17%)
Female	Lung	55_64	198,022	6221 (2530, 9790)	3.14% (1.28%, 4.94%)
Female	Lung	65_74	321,277	8572 (3249, 13738)	2.67% (1.01%, 4.28%)
Female	Lung	75_84	323,392	7030 (2235, 11701)	2.17% (0.69%, 3.62%)
Female	Lung	85+	138,302	1027 (−1501, 3477)	0.74% (−1.09%, 2.51%)
Female		TOTAL	1,980,162	39,767 (9312, 69381)	2.01% (0.47%, 3.5%)
Male	Colon	45_54	29,026	−54 (−661, 531)	−0.19% (−2.28%, 1.83%)
Male	Colon	55_64	60,623	918 (−204, 2004)	1.51% (−0.34%, 3.31%)
Male	Colon	65_74	84,009	2034 (721, 3311)	2.42% (0.86%, 3.94%)
Male	Colon	75_84	91,217	3016 (1613, 4382)	3.31% (1.77%, 4.8%)
Male	Colon	85+	52,196	1160 (225, 2066)	2.22% (0.43%, 3.96%)
Male	Lung	45_54	94,257	2741 (1148, 4286)	2.91% (1.22%, 4.55%)
Male	Lung	55_64	278,785	6691 (2910, 10381)	2.4% (1.04%, 3.72%)
Male	Lung	65_74	432,803	12,421 (7214, 17513)	2.87% (1.67%, 4.05%)
Male	Lung	75_84	385,499	8018 (3741, 12209)	2.08% (0.97%, 3.17%)
Male	Lung	85+	128,757	1179 (−600, 2916)	0.92% (−0.47%, 2.26%)
Male		TOTAL	1,637,172	38,124 (16106, 59600)	2.33% (0.98%, 3.64%)

Among females, the highest attributable fraction for colon cancer was observed in the 85+ age group (2.53%). For lung cancer, the peak occurred in the 55 to 64 age group (3.14%), while for breast cancer, it was also in the 85+ age group (2.51%). In terms of absolute burden, attributable mortality was greatest for lung cancer in the 65–74 age group, with an estimated 8,572 deaths (95% CI: 3,249–13,738) linked to nuclear proximity.

Among males, the highest attributable fraction for colon cancer was observed in the 75–84 age group (3.31%), while for lung cancer, it peaked in the 65–74 age group (2.87%). The greatest attributable mortality was also seen in lung cancer among those aged 65–74, with an estimated 12,421 deaths (95% CI: 7,214–17,513) attributed to nuclear proximity.

Summing across all age groups, the estimated total attributable lung, colon, and breast cancer deaths over the study period were 39,767 (95% CI: 9,312–69,381) for females and 38,124 (95% CI: 16,106–59,600) for males.

Sensitivity analysis

To evaluate the stability and robustness of our findings, we conducted a LKO sensitivity analysis, where we randomly excluded 200 counties (approximately 10% of the dataset) in each iteration and repeated the analysis 100 times for each cancer type, sex, and age group.

Figure 4 presents the distribution of LKO estimates across age groups for breast (females only), colon, and lung cancers, stratified by sex. The results demonstrate that the effect estimates remained consistent across iterations, with narrow interquartile ranges indicating low sensitivity to random county removal. Estimates for all cancer types and both sexes were highly consistent across age groups, reinforcing the reliability of our main findings.

Fig. 4.

LKO sensitivity analysis for nuclear power plants proximity adjusted (Table 2S) effect estimates (proximity coefficient), randomly excluding 200 out of 1922 counties across 100 iterations for each cancer type, sex, and age group.

Additionally, we performed a reactor type-specific sensitivity analysis, where nuclear power plants were categorized into three reactor types: Boiling Water Reactors (BWR), Pressurized Water Reactors (PWR), and all other reactor types. This categorical variable was included as a covariate to assess whether reactor type influenced the observed associations. We found that reactor type was not a significant predictor of cancer mortality (Table 3S in the supplementals), suggesting that the observed associations were not driven by differences in reactor technology.

As a negative-outcome control, we repeated the models using accidental deaths (ICD-10 codes V01–V99, W00–W99, X00–X59, and Y85–Y86). We found no associations between nuclear power plant proximity and accidental mortality across age and sex groups (Supplementary Table 4S), consistent with expectations for a negative control.

DISCUSSION

We examined the association between long-term proximity to nuclear power plants and lung, colon, and breast cancer mortality, utilizing a spatially resolved, inverse-distance weighted proximity metric that captures cumulative proximity (sum of proximities) from multiple plants within 200 km of each U.S. County population center. Our findings indicate positive associations between nuclear power plants proximity and cancer mortality, with stronger effects observed in older age groups (Fig. 2). This pattern is consistent with the age-related increase in cancer risk, which likely reflects cumulative lifetime power plants proximity and age-dependent biological vulnerabilities.

For colon cancer, the highest attributable mortality burden was observed in males aged 75–84 and females aged 85 +. For lung cancer, the greatest attributable mortality occurred in both sexes aged 65–74. Among females, breast cancer showed the highest attributable mortality burden in the 85+ age group. These patterns likely reflect a combination of increased vulnerability among older adults and the underlying demographic structure of the U.S. population (Table 2).

Ionizing radiation is a well-established carcinogen, with extensive epidemiologic evidence linking radiation exposure to an increased risk of cancer [33–39]. Numerous studies have specifically identified lung, colon, and breast cancers as radiation-associated malignancies, demonstrating a direct link between ionizing radiation exposure and cancer development [2, 3, 10, 11, 39–44].

In 2022, breast cancer was the most frequently diagnosed cancer in women worldwide, accounting for approximately 2.3 million new cases and representing about 25% of all cancers among women. Due to improved screening practices and effective treatments in many regions, breast cancer has also become the most prevalent malignancy globally, with an estimated 8.8 million women living who have been diagnosed within the past 5 years [9]. Radiation-induced breast cancer risk is supported by epidemiological evidence primarily derived from medically exposed populations [10].

A study from South Korea examining the impact of proximity to nuclear power plants found that females living near nuclear facilities had an adjusted odds ratio (OR) of 2.61 (0.96–7.06) for developing breast cancer and an adjusted OR of 2.19 (1.25–3.85) for developing radiation-sensitive cancers, including breast cancer [15]. Similarly, a study in Spain assessing cancer risk within 30 km of the El Cabril nuclear facility reported an adjusted relative risk (RR) of 1.476 (1.085–2.007) for breast cancer mortality [16]. In the UK, an investigation into breast cancer mortality near the Bradwell nuclear power station found significant adjusted RRs ranging from 1.7 to 2.2 [45]. These findings are consistent with our highest estimates, in which we observed an RR of 1.16 (1.08–1.26) for breast cancer mortality among females aged 85+ residing within 5 km of a nuclear plant.

Colon cancer is a significant contributor to the global cancer burden, with over 1.1 million new cases diagnosed worldwide in 2022 [9]. Several studies have investigated the potential link between proximity to nuclear facilities and colorectal cancer risk. A study in Spain found an adjusted relative risk (RR) of 1.53 (1.009–1.317) for developing colorectal cancer among individuals living within 15 km of nuclear fuel facilities [16]. Similarly, a study in Canada examining colon and rectum cancer risk within 25 km of the Pickering nuclear plant reported an adjusted standardized incidence ratio (SIR) of 1.41 (1.36–1.45) [1]. Our findings are consistent with these results, showing the highest relative risks of 1.16 (95% CI: 1.07–1.27) among females aged 85 and older, and 1.23 (95% CI: 1.11–1.35) among males aged 75–84 living within 5 km of nuclear power plants.

Lung cancer was the most frequently diagnosed cancer worldwide in 2022, with nearly 2.5 million new cases, accounting for 12.4% of all cancer diagnoses. It remains the leading cause of cancer mortality, responsible for approximately 1.8 million deaths, underscoring its poor prognosis and high fatality rate [9]. Despite well-known associations between lung cancer and smoking, radiation exposure remains an important environmental risk factor, as indicated by studies on ionizing radiation effects on the lungs [40, 46].

In Spain, a study assessing lung cancer risk in relation to proximity to nuclear plants and fuel facilities found adjusted relative risks (RR) of 1.124 (1.015–1.246) for individuals near nuclear fuel facilities and 1.242 (1.056–1.461) for those near the Andújar facility. The study also examined lung cancer risk before and after the startup of nuclear facilities, reporting adjusted RRs of 1.44 (p = 0.0158) near the Garoña facility, 1.355 (p = 0.0173) near the Vandellós facility, and 1.37 (p = 0.0451) near the Ciudad Rodrigo facility [16].

For colon and lung cancers, our results align with established patterns reported in BEIR VII (2006) and EPA (2011, EPA 402-R-11–001) [33, 34]. BEIR VII noted modest sex differences in colon cancer, with slightly higher excess relative risks in men (ERR/Sv ≈ 0.63) than in women (ERR/Sv ≈ 0.43), while both sexes showed clear increases with exposure. Lung cancer demonstrated a more pronounced disparity, with BEIR VII estimating an ERR/Sv of 1.40 in females compared to 0.32 in males, suggesting substantially greater susceptibility among women despite higher smoking-related background risks in men. EPA’s 2011 mortality projections corroborated these sex-specific differences: colon cancer risks were higher in males (ERR-based LAR 74 per 10,000 person-Gy) than in females (45 per 10,000 person-Gy), while lung cancer risks were markedly higher in females (ERR-based LAR 384 per 10,000 person-Gy) than in males (140 per 10,000 person-Gy). Consistent with these reports, our stratified analyses indicated sex-dependent variation, with females exhibiting stronger associations for lung cancer mortality and more modest differences for colon cancer.

In our study, the highest observed relative risk was 1.22 (95% CI: 1.08–1.38) among females aged 55–64 and 1.21 (95% CI: 1.11–1.31) among males aged 65–74 residing within 5 km of a nuclear plant. These findings align with previous research and further strengthen the evidence for a potential association between proximity to nuclear facilities and lung cancer mortality.

Compared to our previous study [19] investigating the association between proximity to nuclear power plants and all-cancer mortality, these three cancer types accounted for an estimated 39,767 deaths (95% CI: 9312–69,381) among females and 38,124 deaths (95% CI: 16,106–59,600) among males, although that study was limited to 2018 and had a smaller counties count due to unavailability of covariates.

Although evidence regarding cancer risks near nuclear facilities remains heterogeneous, findings from nuclear disasters underscore the biological plausibility of radiation-induced malignancies. The Japanese atomic bomb survivor Life Span Study identified elevated risks of leukemia and solid cancers, including colon, lung, and breast cancers, persisting decades after exposure to high doses and dose rates [2, 4, 47]. More recently, studies following the Fukushima accident have reported increased incidence of childhood thyroid cancer [48, 49].

We hypothesize that while variability in findings may partially stem from differences in nuclear technology, emission controls, radiation dose levels, and local demographic characteristics, a substantial portion of this heterogeneity in previous findings is likely due to methodological differences. Variability in exposure assessment approaches, outcome definitions, statistical power, and study design has contributed to mixed results across studies.

Many earlier investigations relied on binary proximity metrics [1, 12–14]. which may have oversimplified exposure distributions. In contrast, our study employs a continuous, inverse-distance weighted metric, providing a more precise representation of long-term cumulative proximity from multiple nuclear facilities. Additionally, our analysis spans 1922 counties over 21 years in the U.S., covering more than 3.5 million cancer mortalities, while incorporating spatial variability and a comprehensive set of covariates to strengthen statistical rigor. Unlike single-facility or geographically limited studies [1, 14]. our national-scale approach significantly enhances statistical power, improving the ability to detect associations and capture proximity gradients more effectively.

Our study has several important limitations. First, the use of aggregated county-level mortality data limits our ability to draw definitive conclusions about individual-level risk, given the ecological study design. Second, our proximity model assumes equal contributions from all nuclear facilities within 200 km, not accounting for variability in plant-specific emissions. Third, the analysis assumes long-term residential stability, potentially overlooking population mobility and changes in proximity over time. Lastly, while we stratified analyses by cancer type, age, and sex, the possibility of residual confounding from lifestyle factors or other environmental exposures cannot be ruled out.

Despite these limitations, our research represents the first comprehensive national assessment of cancer-specific mortality risks related to proximity to nuclear power plants in the United States. By incorporating a continuous exposure assessment approach and nationwide mortality data, we provide critical evidence addressing longstanding epidemiological uncertainties. These findings emphasize the importance of continued monitoring and targeted public health interventions near nuclear facilities, particularly among older populations who exhibit increased vulnerability to radiation-associated malignancies.

CONCLUSION

We observed elevated mortality from breast, colon, and lung cancers in U.S. counties with greater cumulative proximity to nuclear power plants, with the strongest associations found in older adults. These findings demonstrate that residential proximity to nuclear facilities is associated with, and suggest that it may contribute to, cancer-specific mortality, underscoring the need for continued epidemiological monitoring and more precise characterization of exposure pathways.

Given the increasing global reliance on nuclear power as a low-carbon energy source compared to fossil fuels, our findings should incentivize the next level of investigation, drawing attention to the importance of exposure-specific pathways, plant-level emission characterizations, and dosimetry-based assessments. Such studies can advance understanding of long-term and cumulative risks and ultimately provide an evidence base that informs both future research directions and policy considerations.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41370-026-00922-2.

DATA AVAILABILITY

The cancer mortality data utilized in this study was obtained from the Centers for Disease Control and Prevention (CDC) and is confidential; therefore, it cannot be shared. All other data sources are publicly available. Sample code for the statistical epidemiological models used in our analysis can be provided upon request.