A Harvard methodology that claims nuclear power plants cause cancer deaths has been turned against itself, with researchers finding that living near Costco, private colleges, and even Major League Baseball fields can produce far more “attributable” cancer deaths than the original studies found for reactors.
The results are deliberately absurd. Costco reportedly causes over 2.2 million cancer deaths per year, more than 20% of all cancer deaths between 2000 and 2018. Private four-year universities account for 1.59 million deaths. Harvard itself is responsible for 5,842 deaths. Living near a baseball field causes almost 19 times as many cancer deaths as living near a soccer pitch.
These figures come from researchers Adam Stein and his team, who spent months replicating the methodology behind a series of Harvard studies. Their conclusion is blunt: the approach proves nothing about cancer incidence.
The Nuclear Studies
The original research was led by Yazan Alwadi at Harvard’s T.H. Chan School of Public Health. His team published a paper in Environmental Health in December 2025, claiming cancer incidence rose for people living closer to nuclear power plants in Massachusetts. A second nationwide study followed in March in Nature Communications, looking at cancer mortality rates rather than incidence. Additional papers examined lung, breast, and colon cancers in the Journal of Exposure Science & Environmental Epidemiology, and a study of total mortality appeared in Environmental Epidemiology.
The studies build a “proximity score” based on distance from nuclear plants, up to 120 km in Massachusetts and 200 km nationally. Every ZIP code or county within those radii is treated as exposed, with closer locations receiving higher weights. A county near one plant can have a smaller score than a county further from any single plant but within range of many.
From that score, the authors run regressions testing outcomes on proximity and a collection of covariates. They then compute the Attributable Fraction, which derives the number of deaths attributed to the plants from 2000 to 2018.
Why Proximity Isn’t Exposure
Stein and his team argue that proximity is not the same as exposure. There are well-established methods of measuring radiation exposure for nuclear workers, and most closely monitored US workers receive no measurable annual dose. Even when workers are exposed, the average dose is only 2 percent of the occupational limit.
If operators and workers on-site receive an annual dose between zero and one-fiftieth of the occupational limit, how could residents 5, 10, 25, 50, 120, or 200 kilometers away receive any measurable dose from the same plant?
The Replication Effort
The authors supplied eight lines of code and answered questions about the covariates. That did not replicate the results. Most replication was done through first principles combined with trial and error.
Once the team matched the first national study on cancer mortality, they took the methodology and applied it to other landmarks. The results held: everything produces cancer.
The Hedge
Alwadi and his team describe their papers as ecological studies that show association, not causality. Ecological studies look at relationships between outcomes and exposures at a population level.
That framing lets the authors explain the broader world with wide latitude. The use of an “ecological study” allows a lot of leeway in terms of exposure pathways and mechanisms, Stein’s team argues.
The Ridiculous List
Stein’s team compiled a list of absurd findings from applying the methodology:
- Costco: over 2.2 million cancer deaths per year, more than 20% of all cancer deaths between 2000 and 2018
- Private four-year universities: 1.59 million deaths
- Harvard itself: 5,842 deaths
- Living near a baseball field: almost 19 times as many cancer deaths as living near a soccer pitch
The comparison is stark. The methodology designed to prove nuclear plants cause cancer deaths now proves that Costco causes vastly more deaths.
The Claims Against Nuclear Plants
The papers claim that living closer to nuclear power plants increases cancer incidence and mortality. The methodology constructs a proximity score based on distance, with ZIP codes and counties receiving higher weights the closer they are to a plant. If a location is within range of multiple sites, the effect is cumulative.
The authors use statistical coefficients to calculate the Relative Risk for each county, age, and sex group, then compute the Attributable Fraction from which they derive the number of deaths from 2000 to 2018. They state that their methodology provides the scientific basis for an expanded research program.
What The Authors Say
The authors state that their methodology and results provide the scientific basis and public health justification for an expanded research program. Stein’s team disagrees. The studies simply do not prove anything about cancer incidence.
The methodology creates an association between any physical landmark and cancer, then calculates how much cancer is attributable to that thing. That is not the same as proving causality.
The Bottom Line
The methodology is a hammer looking for nails. It finds cancer everywhere, including places that have no plausible mechanism for causing it.
The authors hedge that their papers show association, not causality. That hedge is meaningless when the methodology itself can produce any result you want.
Stein’s team has shown that the Harvard approach is a broken tool. It produces absurd conclusions when applied to familiar, harmless institutions. The methodology does not distinguish between a nuclear plant and a warehouse club.
The takeaway is simple. The Harvard studies do not prove their claims. They prove that proximity can be made to mean anything.
Key facts box:
– Costco: over 2.2 million cancer deaths per year, more than 20% of all cancer deaths between 2000 and 2018
– Private four-year universities: 1.59 million deaths
– Harvard itself: 5,842 deaths
– Baseball vs. soccer: almost 19 times as many cancer deaths from baseball
| Landmark | Deaths attributed |
|---|---|
| Costco | Over 2.2 million |
| Private four-year universities | 1.59 million |
| Harvard | 5,842 |
| Baseball | Almost 19x more than soccer |
This is a story about a methodology that cannot distinguish between a deadly hazard and a bulk spice aisle. It is a story about science being used to prove whatever its users want it to prove. And it is a story about the absurdity of treating proximity as proof of anything at all.
Source material: “Things that apparently cause cancer,” breakthroughjournal.org.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.