Things that apparently cause cancer

Statistical association is not evidence of causation, and this critique reveals how flawed methodologies can generate misleading health statistics.

Researchers have challenged a Harvard School of Public Health methodology used in several studies claiming nuclear power plants cause cancer. The original studies, published between December 2025 and recently, used a "proximity score" based on distance to nuclear facilities, extrapolating cancer incidence and mortality rates from ZIP codes or counties within a 120-200 km radius. The authors of the critique replicated the analysis and applied the same methodology to various other landmarks, including Costco warehouses, private colleges, MLB fields, and state capitals. Their findings suggest that, using the Harvard methodology, proximity to almost any landmark, such as Costco, can be statistically associated with a significant number of cancer deaths. For instance, proximity to Costco was linked to over 2.2 million cancer deaths, and proximity to private four-year colleges was associated with 1.59 million deaths, far exceeding the attributed deaths from nuclear power plants in the original studies. The critics argue that the methodology conflates correlation with causation and does not account for actual exposure pathways.

The core issue lies in the Harvard methodology's failure to establish a causal link between proximity and cancer. By treating proximity as a proxy for exposure without considering actual radiation levels or biological mechanisms, the approach is fundamentally flawed. The critique demonstrates that the statistical model is so generalized that it produces significant, albeit absurd, cancer associations with virtually any physical landmark, indicating that the methodology itself, rather than the landmarks, is what generates the "results."

This critique has significant implications for public trust in scientific research and policy-making based on such studies. It highlights the critical need for rigorous validation of research methodologies, especially in health and environmental science, and suggests that previous findings linking nuclear power to cancer incidence and mortality may be unsubstantiated. Future research must prioritize establishing concrete exposure pathways and causal mechanisms rather than relying on broad correlational data.