You may have recently come across an article like this recently, which breathlessly repeats a press release issued by the University of North Carolina about a study
linking saturated fat and aggressive prostate cancer.
The findings themselves, which were presented on April 18 at the American Association for Cancer Research Annual Meeting, are perfectly sound but are nowhere near as definitive as the press release and any subsequent media attention would have you believe. This highlights a recurrent issue in science reporting that can lead to a lot of misled readers.
The study, with lead author Emma Allot, “Saturated fat intake and prostate cancer aggressiveness: Results from the population-based North Carolina-Louisiana prostate cancer project”, can be found on page 1,048 of the published AARC 2016 abstracts as number 1,760. The collection of abstracts can be obtained for free here at Google Play.
The Allot Study on Prostate Cancer and Saturated Fat: Summary
- 1, 854 men who participated in the North Carolina-Louisiana Prostate Cancer Project
- The level of aggressiveness of the men’s prostate cancers was classified and 321 were considered to have high aggression
- The men were asked about their dietary habits and other risk factors at the time of diagnoses as part of the Project, and this data was used by the researchers
- Various analysis’ were conducted to compare rates of total fat, monosaturated, polyunsaturated, saturated, trans fat, cholesterol intake, and the use of statins (cholesterol-lowering drugs) between the high and mid-to-low aggression groups
- A higher likelihood for being in the aggressive prostate cancer group was found for those with higher saturated fat intakes
- The saturated fat-aggression association was weaker if the person was also taking statins
- There was a trend towards higher cholesterol intake and aggressive prostate cancer having an association, but this did not reach statistical significance
- The impact of saturated fat on cholesterol levels was theorized as a possible mechanism to explain these findings
Where The Reporting Fails
One key difference that can be seen when reading the actual abstract versus the media report is that the authors never use the word “link”. This is an important distinction since, when dealing with scientific studies, saying there is a “link” implies a certain level of evidentiary support. However, that level of evidence is definitely not present in the study for one simple reason: it is preliminary.
Preliminary studies are part of the research process but are an extremely early stage that can be best likened to “proof of concept” with a very loose requirement for “proof”. A preliminary study is used to see if there is any possible phenomenon or bit of data that is worth attempting to conduct more thorough research on; by definition, a preliminary study is incapable of proving a substantial link between anything. At best, a preliminary study can only say, “these two bits of data seem to line up, we should see if that means anything”.
While it can be tempting to describe preliminary results in terms of that a connection “may” exist, to do even that is also highly premature. Preliminary findings are often obtained without adequately accounting for complicating factors or many sources of error. Trying to describe preliminary findings as saying a link “may” exist is kind of like saying a movie “may” be good if all you have seen are the rough jots of a script writer before any brainstorming or draft writing takes place.
Issues With the Allott Study
There is nothing inherently flawed about the study conducted by Allott and her fellow researchers. Data was collected, data was reported, and preliminary associations that could be researched further were pointed out. However, due to the importance in emphasizing why there should not be reporting (press release or otherwise) on preliminary findings, here are some of the immediate problems with trying to interpret the findings as providing a “link” between saturated fat and prostate cancer aggressiveness:
- Cancer does not appear overnight. Without knowing when these men developed prostate cancer (not just when it was diagnosed) it is impossible to even know the timeframe in which their diets would have been relevant.
- The subset of men with highly-aggressive cancer was very small compared to the mid-to-low aggression groups. This means that any apparent trends in the highly aggressive group is going to appear much larger than a bigger group might show (greater risk of false positives)
- The mention of a “trend” in cholesterol levels is a way of drawing attention to something the authors wants to be relevant but knows the data doesn’t support (it doesn’t meet statistical significance)
- Statins were found to be associated with mitigating aggression associations but cholesterol levels themselves (which are what statins affect) didn’t show significant positive or negative associations with prostate cancer aggressiveness. How can statins have a significant effect if the thing it changes doesn’t show a relation to cancer aggression? This suggests at least one of the values to be the result of statistical noise.
None of this is meant to detract from Allott and her group’s work. It is merely to point out some of the reasons why trying to claim a “link” from preliminary research is an inherently flawed activity.
- The study is fine as a piece of preliminary research but it is far too early to try and draw meaningful links or conclusions
- There are too many possible confounding factors and limited rigor to try and derive links, conclusions, or anything that really warrants media interest or the attention of the public without some very large disclaimers
- Pay attention to the context of the study itself, not the way it gets reported
Allott, E., et. al., “Saturated fat intake and prostate cancer aggressiveness: results from the population-based North Carolina-Louisiana prostate cancer project,” Proceedings of the AACR Annual Meeting, 2016, page 1760; https://play.google.com/store/books/details/
“Increased saturated fat intake linked to aggressive prostate cancer,” Science Daily web site, April 19, 2016; https://www.sciencedaily.com/releases/2016/04/160419081941.htm, last accessed April 20, 2016.