Since my post earlier today on the red meat study, there was a comment with a link to an article critical of the study, and a kind reader emailed me the actual journal article.
The article linked to above points out the most serious problem with the study—that red meat causes a similar increase in all types of deaths. The study tracked four types of deaths: (1) cancer; (2) cardiovascular disease; (3) injuries and sudden deaths; and (4) all other deaths. For men, deaths from all four causes increased with red meat consumption, but for some strange reason, women who ate more red meat were less likely to die from injuries and sudden death, but there weren’t enough deaths in this category for the result to be statistically significant.
The mechanism by which meat might cause death by cardiovascular disease is that the cholesterol in the meat causes atherosclerosis which causes heart disease. But why does meat also cause death by cancer, all other diseases, and even accidents? Is meat just such a vile substance? Or is low meat consumption indicative of someone from a higher social class and who has access to better healthcare, and has higher future time orientation and therefore is less likely to partake in risky behavior?
The study itself claims to have accounted for confounding factors, including education, marital status, BMI, smoking history, alcohol intake, vitamin supplement use, fruit consumption, and vegetable consumption. It should be noted that all of these factors are related to each other. According to the data printed in the journal article, people who ate less red meat were less likely to smoke, had more education, were more likely to exercise, and ate more fruits. Important confounding factors related to social class which are missing include income, wealth, and IQ. And all college graduates are lumped together, so someone with a graduate degree from Harvard is considered to be in the same category as someone with a bachelor’s degree from a public commuter college obtained at the age of 35.
If you’ve played around with a lot of multiple regression analyses, you would know that sometimes, when you throw in a lot of variables that are correlated with each other, you get some weird results. For all we know, the particular confounding factors tracked in the study where cherry picked to achieve the desired results while creating the illusion that confounding factors were considered.
Recent Comments