Read Original Story

Vegetarians who ate fish had lowest colorectal cancer risk, study says


2 Star

Vegetarians who ate fish had lowest colorectal cancer risk, study says

Our Review Summary

The story summarizes a recent paper in JAMA Internal Medicine that reports a vegetarian diet is associated with a reduced risk of colorectal cancer. The story goes into some detail on the decreased risk for various types of vegetarian diets, and discusses some of the uncertainty regarding what may be responsible for the correlation between diet and cancer risk. However, the story never gives readers an accurate breakdown of the primary limitations of the study, and the statistics used in the story will give readers an inflated sense of the benefits. Not only is the reduction in cancer risk very small in absolute terms, but many of the results mentioned were not statistically significant. While the traditional definition of statistical significance has some problems, as we’ve discussed recently on the blog, we’d still expect stories to alert readers when findings do not achieve this benchmark.


Why This Matters

As the story notes, colorectal cancer is the second leading cause of cancer death in the United States. The CDC also reports that colon and rectal cancers are the third most common cancer in men and in women. Given that more than 130,000 people are diagnosed with colorectal cancer, and more than 50,000 die of colorectal cancer, each year in the U.S., this is clearly an important issue — and it is well worth discussing measures that can lower the risk of contracting the disease.


Does the story adequately discuss the costs of the intervention?

Not Applicable

Dietary changes can affect budgets. But the general dietary changes discussed in this story, and in the study it is based on, don’t lend themselves to an analysis of costs. We’ll rate this as not applicable.

Does the story adequately quantify the benefits of the treatment/test/product/procedure?

Not Satisfactory

The story uses only relative risk reductions to describe the impact of vegetarian diets, which provides readers with an exaggerated sense of the overall size of the benefit.

Let’s look at what is said in the story: “…researchers from Loma Linda University found that vegetarians were 22% less likely than meat-eaters to be diagnosed with colon cancer or rectal cancer.” But here are the absolute numbers from the published article: During the 7.3 year follow up period, a total of 252 cases of colorectal cancer were identified in the 40,367 vegetarians studied (62.4 cases per 10,000) and 238 cases in the 37,292 non-vegetarians studied (63.8 cases per 10,000). That is a difference of 1.4 cases per 10,000 subjects. The relative reduction in risk does not play out to be very large in absolute numbers.

And again the story says: “Lacto-ovo vegetarians (who eat eggs and dairy foods but not fish or other meats) came in second, with an 18% reduced incidence of colorectal cancer. They were followed closely by vegans (who don’t eat eggs, dairy products or any kind of meat), who were 16% less likely to get a colorectal cancer diagnosis.” But when we look at the actual study, we find that the confidence interval in both of these cases crosses 1, which means that the findings are not statistically significant (and would certainly be very small in absolute terms).

When rolled up together, there does appear to be a slightly smaller risk of colorectal cancer associated with a vegetarian diet compared with a non-vegetarian one.  Having said that, when you drill down into the different sub-diets, only the pesco-vegetarian diet was statistically significant on its own.  None of the other diets are associated with a reduce risk of either colon or rectal cancer. Compared with the relative risks touted in the story, the absolute numbers derived from the study provide a somewhat different and less enthusiastic picture of the benefits.

Does the story adequately explain/quantify the harms of the intervention?

Not Applicable

There are really no harms to discuss when addressing these dietary changes, particularly given the extremely broad definitions of the vegetarian diet in the relevant study. Not applicable.

Does the story seem to grasp the quality of the evidence?

Not Satisfactory

The story discusses the size and length of the study, which is good. It also tells us that all of the relevant “risk-reduction calculations were adjusted to account for age, gender, race, education level, daily calorie intake, exercise habits, smoking and drinking behavior, family history of colorectal cancer, and other medical conditions.” That’s an important note, and the story also links directly to the paper in JAMA Internal Medicine — which is always a plus.

However, the story doesn’t tell readers that this was an observational study — nor does it fully discuss the limitations of such a study. For example, while the story does note that “it will take more research to determine whether the pesco-vegetarians were better off because they ate fish,” the story does not make it clear to readers that the study shows only a correlation between diet and cancer risk — not a cause-and-effect relationship. That’s a critical caveat. We found an example of the language we were looking for in a competing HealthDay story on the same study. The story quoted Dr. Alfred Neugut, professor of epidemiology at Columbia University Medical Center in New York City, who shared the following insights:

Dietary studies can only show an association between cancer and diet, not a cause-and-effect relationship, Neugut said. “That’s the problem in dietary studies of cancer. We don’t know exactly what the connection is,” he said.

Neugut said that a vegetarian diet might be a sign of other healthy behaviors, such as exercising and not smoking, which could also reduce the risk for cancer.

Only two sentences of extra text, but a big impact on readers’ interpretation of the study.

Does the story commit disease-mongering?

Not Satisfactory

The first twelve words of the story are: “Colorectal cancers kill more Americans than any other cancer except lung cancer.” But that line doesn’t give readers any context. In short, readers might not know what that actually means. For example, the CDC reports that more than 51,000 people died of colorectal cancer in 2011. The CDC also reports that a 30-year-old man has a 0.97 percent chance of contracting colorectal cancer over the next 30 years. That kind of information would have been easy to include, and may have helped readers understand their colorectal cancer risk in meaningful terms.

Does the story use independent sources and identify conflicts of interest?

Not Satisfactory

There are no apparent conflicts of interest, but the story does not include input from any independent sources.

Does the story compare the new approach with existing alternatives?

Not Satisfactory

The story does not discuss other interventions that can reduce the risk of colorectal cancer, such as screening either by colonoscopy or fecal occult blood testing.

The story does mention regular exercise, not smoking, and limiting alcohol intake, but only to the extent that those are all things that are recommended by the Seventh Day Adventist church. The story would have been better if it had made an explicit link between those behaviors and cancer risk.

Does the story establish the availability of the treatment/test/product/procedure?

Not Applicable

Not applicable. Vegetarian foods are widely available.

Does the story establish the true novelty of the approach?


The story states that previous studies have explored correlations between diet and colorectal cancer risk, noting that “it’s not exactly surprising that vegetarians fared better than meat-eaters in this study.”

Does the story appear to rely solely or largely on a news release?


The story does draw from a JAMA news release, but the reporter clearly dug into the paper for additional information. Still, we would prefer if the story had included input from researchers with relevant expertise who were not affiliated with the paper.

Total Score: 2 of 7 Satisfactory


Please note, comments are no longer published through this website. All previously made comments are still archived and available for viewing through select posts.