Plant-Based Diets and Cancer Risk: What the Evidence Shows

In 2015 the World Health Organization's cancer agency classified processed meat as a Group 1 carcinogen and red meat as a probable carcinogen, Group 2A (IARC 2015). The headlines wrote themselves: bacon as dangerous as cigarettes. That reading is wrong, and the reason it is wrong is the single most useful idea on this page.

Hazard is not the same as risk

The International Agency for Research on Cancer sorts substances by how strong the evidence is that they can cause cancer at all. That is a statement about hazard, meaning the confidence that something is capable of causing cancer. It is not a statement about risk, meaning how much your chances actually go up, or by how much, when you eat a given amount.

Group 1 means scientists are confident the link is real. It does not mean the effect is large. Tobacco smoking is Group 1. So is processed meat. So is sunlight, alcohol, and salted fish. They share a category because the evidence that each can cause some cancer is solid, not because they raise your risk by the same amount. Smoking raises lung cancer risk many times over; processed meat nudges colorectal cancer risk up by a much smaller fraction. The classification grades the strength of the evidence, not the size of the effect (IARC 2015).

This is exactly the kind of distinction the how to read nutrition studies page is built around, and it is worth slowing down for, because the next number gets misread for the same reason.

Relative risk is not the same as absolute risk

The often-cited figure is that eating about 50 grams of processed meat a day, roughly one hot dog or a couple of slices of bacon, is associated with about an 18% higher risk of colorectal cancer (IARC 2015). Eighteen percent sounds alarming until you ask: eighteen percent of what?

That 18% is a relative increase. It tells you how much a baseline risk goes up in proportion, not the actual number of extra cases. To know whether 18% is a lot in practice, you need the absolute risk, the baseline it is multiplying.

Colorectal cancer is reasonably common but still affects a minority of people over a lifetime. Take a rough illustrative baseline: suppose about 5 in 100 people would develop colorectal cancer without the extra processed meat. An 18% relative increase on that baseline raises it to roughly 5.9 in 100, an absolute increase of just under one extra case per 100 people. (The exact baseline varies by country, age, sex, and screening, so treat the specific numbers as an illustration of the arithmetic rather than a precise personal forecast.) A modest relative increase applied to a moderate baseline produces a small absolute change. That is the honest shape of the processed-meat finding: the evidence it does something is strong; the something it does is small per person, though it adds up across a whole population eating processed meat daily.

None of this makes the classification meaningless. A small per-person risk spread across hundreds of millions of regular consumers translates into a real number of cancers at the population level, which is why public-health bodies bother to flag it. The mistake is jumping from "Group 1" to "as deadly as smoking for any individual." The category is about certainty. The size of the effect is a separate question, and for processed meat the effect is modest.

What the broader evidence shows

Step back from the single most quoted statistic and the wider pattern across diet research is consistent and, on the whole, more encouraging than the meat headlines suggest. It points toward what people eat more of, not only what they eat less of.

A dose-response meta-analysis pooling many prospective studies found that higher intakes of whole grains, vegetables, fruit, nuts, and legumes track with lower all-cause mortality, while higher red and processed meat intake tracks with higher mortality (Schwingshackl et al. 2017). "Dose-response" matters here: the more of the protective foods people ate, the lower the associated mortality, and the relationship moved smoothly rather than flipping at a single threshold. That kind of gradient is one of the features that makes an association more credible, though it still cannot by itself prove the food is the cause.

Fiber, which comes almost entirely from whole plant foods, shows up strongly. A series of systematic reviews and meta-analyses commissioned by the WHO found that people eating the most fiber had roughly 15 to 30% lower all-cause mortality and lower incidence of heart disease, stroke, type 2 diabetes, and colorectal cancer than those eating the least (Reynolds et al. 2019). Colorectal cancer, the same outcome processed meat is tied to, appears on both lists: more fiber with less of it, more processed meat with more of it. Those percentages are again relative reductions, so the same caution applies in reverse; a 15 to 30% lower relative risk is meaningful but is not a guarantee for any one person.

At the global scale, the Global Burden of Disease analysis of diet estimated that the largest dietary contributors to death and disability worldwide are diets low in whole grains, fruits, nuts, and vegetables, alongside high sodium intake (GBD 2017 Diet Collaborators 2019). Notably, the biggest diet-related burden in that analysis comes from the protective foods people are missing, not solely from the harmful foods they eat. The practical reading is that adding plants does at least as much work as subtracting meat.

The honest caveats

Almost all of this evidence is observational. Researchers follow large groups of people, record what they eat, and watch what happens to their health over years. That design can establish strong, repeated associations, and several of these findings show the dose-response gradients that make a causal story more plausible. It cannot, on its own, prove that the food caused the outcome.

The main reason for caution is confounding. People who eat a lot of processed meat tend to differ from people who eat a lot of vegetables in other ways: smoking, drinking, exercise, body weight, income, and access to care. Good studies adjust statistically for the factors they can measure, but no adjustment is perfect, and some confounders go unmeasured. So the careful phrasing throughout is "associated with," not "causes."

What the evidence supports is measured. Processed meat is a confirmed carcinogen with a small per-serving effect on colorectal cancer risk. Diets built on whole plant foods are associated with lower risk of several outcomes, including some cancers, in large observational studies. A plant-based diet is not a treatment for cancer and will not prevent it in any individual case. What the data describe is a shift in the odds across a population, which is a real and worthwhile thing, and also a more modest claim than the one the original bacon-versus-cigarettes headline implied.