My title is literal. It seems that this study has more fervor than the last bout of ‘meat gonna kill you studies’ and everyone is throwing their hats in the ring on this one. I guess its my shot to beat a dead horse.
Given how a plethora of other blogs have already gotten around to dissecting the study, in either a good or bad fashion (in regards to what they chose to attack, and how well they did it), I’m just going to make this blog post in a pseudo-Q&A format with quotes. The quoted are questions I have seen asked frequently, and between them are my thoughts.
What do you think about the new Meat will kill you study?
Its a very good study. It controlled for a lot of variables that past studies did not, and found that meat was confidently but weakly associated with cancer and mortality
So meat is going to kill you?
No, its still an association. The association is getting stronger with each passing study, but the study design doesn’t allow for causality to be stated.
Its becoming more and more likely that eating red meat, rather than other factors, is associated with an increase in mortality.
But correlation does not equal…
I’m going to stop you before you say ‘causation’ and ask you to shut the fuck up.
Correlation equals correlation, causation equals causation. This study didn’t go out and try to find causation though, it went out to find correlation and did a good job of it.
But the Food Frequency questionnaire used has a reproducability in the 0.5-0.6 range, and it relies on superhuman recall. Truly this is shit!
Do you fail to understand how epidemiological research is done? The former is pretty good and the latter conjecture for epidemiology. Sure they’re shit if you compare them relative to intervention studies, but then why don’t you go to an intervention study and critique their sample sizes for only being 500 when epidemiology routinely surveys over 50,000?
(Here’s the survey btw)
All studies have their strengths and weaknesses, and some things cannot be compared. Epidemiology is great in regards to sample size and external validity (what is seen is most likely indicative of the population as a whole) but they must trade off internal validity in order to achieve that (where intervention research shines, being able to place causation on a certain variable).
If you micromanage your diet and control as many variables as possible, then causation is probably what you want. Look for internal validity and interventions, but don’t knock down on external validity; it just doesn’t apply to you.
So is this study good?
Its good survey research. Pretty shit intervention if you want to judge it on those critera, but good if you judge it on the criteria that it is supposed to be judged on. Perhaps you remember this quote?
“Everybody is a genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is stupid.” – Einstein
“All forms of research are legitimate, but if you judge a survey on its ability to find a causative variable then of course its going to look like shit.” – Silverhydra
What conclusions can I draw from this study?
This study shouldn’t change your life too much; you are not a city nor a government organization.
We know that red meat is associated with mortality and cancer; this is either due to red meat itself or another variable not included in the large list that the authors controlled for (WeightyMatters has them in bullet form). So if a large regulatory body wants to act to reduce cancer and mortality rates, then advising against red meat is probably a good idea.
Does red meat cause cancer? It is not known. If I had to guess, I would assume that it may contribute to cancer risk but could possibly be negated by coingested foods or the state of an overall diet (this tends to be a motif with diet); that is conjecture though.
I would be interested in seeing how much PolyAromatic Hydrocarbons contribute (read up about them here) merely because although they should not differ between meats, red meats tend to have more flame-grill affinity than does chicken or fish; possibly more PAHs.
Would actively advising against red meat, on a population wide basis (note: this means screw your personal life and opinions, but look at the collective whole) reduce cancer rates? Probably. If everybody reduced red meat intake *and* it was causative, then there is an obvious reduction. If everybody reduced red meat intake and inadvertently reduced partaking in whatever the hell was associated with it, still a reduction.
Although the above is definitely my sentiment and is the attitude one should take towards epidemiology, it may not apply so much in this particular instance. In controlling for all those variables, the red meat group only have an 0.2-fold increased change of death. Its a much higher risk if you didn’t control for smoking (cancer sticks + cancer meat toxins == bad), but if anything this study showed that the overall association between red meat excluding other stuff is quite small.
Why did you decide to write this post?
It seems people have started to really appreciate intervention research, which is great. Sadly, this has occurred alongside less appreciation for survey/epidemiological research as well as thinking intervention research is infallible (unless it goes against your beliefs of course!). All research has positives and negatives.
A lot of people, both bloggers and commentors, are trying to hold this study up to the standards of intervention research.
Is it too much to state that this study does not apply to the individual, should not concern the individual, and is worthless in a repertoire of studies that should dictate diet micromanagement?
Of course, by talking about it in the media to the uneducated hordes or persons who are afraid of cancer risk you effectively cater to a population. Its not weird to assume that people who change their dietary patterns because of this study are also avoiding whatever association is causing the increase in mortality; these are the same people who are more likely to have bad lifestyles overall rather than eating a nicely sauteed steak on a bed of mixed veggies after coming back from the gym with their paleo spouses and partaking in the newest research on supplementation. We’re sort of freaks in eyes of society anyways; healthy freaks, but freaks nonetheless.
TOO LONG DIDN’T READ?
For all of those citing about how shit survey research can be, realize that interventions can be shit as well; watch this, it kind of attacks all science and exemplifies that you can only apply things in the context of the experimental method. Survey research should be kept on the level of the society, intervention on the level of personal interventions. I defer to Ben Goldacre as he’s a lot better at speaking than I am at talking.
Epidemiology is the science of how we know, in the real world, if something is good for you or bad for you.
Circlejerking, Micromanaging subcultures be damned.
I’ll probably write a more comprehensible post in the future about Internal v. External ‘truths’ in science, but at the time being just be aware that this type of research has its place. Its not something individuals should be worried about, but don’t disrespect it.