Residential proximity to nuclear power plants has been an ongoing public health concern, but epidemiological evidence remains inconsistent. This nationwide ecological study aimed to assess whether residential proximity to operating nuclear power plant units is associated with district-level cancer incidence across South Korea between 1999 and 2018.
The analysis covered 123 administrative districts in South Korea for the period 1999–2018. District-level age-standardized incidence rates (ASRs) for total and site-specific cancers were used as the outcome measures. ASRs were log-transformed before modeling to stabilize variance and meet modeling assumptions.
Residential proximity to nuclear power plants was estimated with an inverse-distance-weighted index. This metric incorporated contributions from multiple operating reactor units rather than relying on distance to a single facility. To account for potential latency between exposure and cancer diagnosis, the authors applied moving-average windows when constructing the exposure metric.
Gaussian generalized estimating equation (GEE) regression models with an identity link were fitted using the log-transformed district-level ASRs as outcomes. Models adjusted for a range of district-level covariates covering environmental, socioeconomic, and demographic factors. The GEE framework accounts for correlation in repeated measures over time at the district level. The abstract does not report the full set of covariates or specific modeling diagnostics; those details would be in the full text.
The study reported a positive association between nuclear power plant proximity and total cancer incidence in both males and females at the district level.
In males: positive associations were observed for total cancer and liver cancer across all moving-average windows considered. Lung cancer also showed positive associations in most moving-average windows, but not in the 25-year moving-average window.
In females: positive associations across all moving-average windows were reported for total cancer, colorectal cancer, and thyroid cancer.
The abstract does not provide numerical effect estimates, confidence intervals, p-values, or model coefficients for these associations; such numerical results are not reported in the source abstract and therefore are not reproduced here.
The observed positive associations between the inverse-distance-weighted proximity index and several cancer outcomes suggest a potential link between residential proximity to operating nuclear power plants and increased cancer incidence at the population level in South Korea. The use of an index that weights nearby reactors and the application of moving averages to represent latency strengthens the exposure characterization for an ecological analysis.
However, ecological designs assess associations at the group (district) level and cannot establish individual-level causation. The authors state these results provide a basis for further population-based research to clarify long-term associations between residential proximity to nuclear power plants and cancer incidence.
Strengths described or implied by the abstract:
Nationwide coverage across 123 administrative districts over a 20-year period (1999–2018) supports broad geographic and temporal representation.
Use of an inverse-distance-weighted proximity index that integrates contributions from multiple reactor units offers a more nuanced exposure metric than nearest-facility distance alone.
Application of moving-average exposure windows to address latency is appropriate for cancer outcomes with potentially long induction periods.
Limitations inherent to the study design, as inferred from the abstract:
The ecological design limits causal inference at the individual level and is subject to ecological fallacy.
The abstract does not present effect sizes, uncertainty estimates, or sensitivity analyses; these details are not reported in the source abstract.
Potential for residual confounding remains despite adjustment for environmental, socioeconomic, and demographic covariates; the abstract does not specify which covariates were included.
Exposure misclassification at the individual level is possible because district-level proximity indices may not capture individual residential histories, mobility, or indoor/outdoor exposure differences.
In this ecological study covering 123 South Korean districts from 1999 to 2018, residential proximity to nuclear power plants—measured using an inverse-distance-weighted index with moving-average latency windows—was positively associated with district-level total cancer incidence in both sexes and with several site-specific cancers (male liver and lung; female colorectal and thyroid) across most latency windows. The authors conclude these findings warrant continued population-based research to better characterize long-term associations between living near nuclear power plants and cancer incidence.
Note: The abstract did not report numerical effect estimates, confidence intervals, model coefficients, or the complete list of covariates; those details were not available in the provided source abstract.