Red meat gives women strokes? No.

Source: Edmonton Journal

Chris Sturdy emailed me a link to a news article claiming "Eating lots of red meat ups women's stroke risk."   I decided to blog about it because it illustrates the difference between relative and absolute risks, and probable investigator bias or poor study design.  The article states:

"Dr. Susanna Larsson of the Karolinska Institutet in Stockholm and her colleagues looked at 34,670 women 39 to 73 years old. All were free of cardiovascular disease and cancer at the beginning of the study, in 1997."

It goes on:

"During 10 years of follow-up, 1,680 of the women (4 per cent) had a stroke....

"When the researchers divided women into five groups based on how much red meat they reported eating, they found that those in the top fifth, who ate at least 86 grams daily (3 ounces) were at 22 per cent greater risk of cerebral infarction than women in the bottom fifth (less than 36.5 grams, or 1.3 ounces, daily).

"Women who ate the most processed meat (at least 41.3 grams, or 1.5 ounces, a day) were at 24 per cent greater risk of this type of stroke than women who consumed the least (less than 12.1 grams, or less than half an ounce a day)."

From the first sentence in the passage immediately above, we know that in this study, 96% of the women did not have a stroke.  If we go to the abstract of the original study, "Red meat consumption and risk of stroke in Swedish women, we find that the so-called 22% greater risk was calculated on a relative rather than absolute basis.

"During a mean follow-up of 10.4 years, we ascertained 1680 incident cases of stroke, comprising 1310 cerebral infarction, 154 intracerebral hemorrhage, 79 subarachnoid hemorrhage, and 137 unspecified stroke. Total red meat and processed meat consumption was associated with a statistically significant increased risk of cerebral infarction, but not of total stroke, intracerebral hemorrhage, or subarachnoid hemorrhage. The multivariable RR of cerebral infarction for the highest versus the lowest quintile of consumption were 1.22 (95% CI, 1.01–1.46) for red meat and 1.24 (95% CI, 1.04–1.49) for processed meat. Fresh (unprocessed) meat consumption was not associated with total stroke or with any stroke subtype."

 

To clarify, if 4 of every 100 people had a stroke (as in this study), the absolute risk for stroke was 4%.  If in one of the subgroups 5 of every 100 people had a stroke, their absolute risk was 5%.  But if you compare the two groups, the subgroup had a 25% greater risk than the other, because 5% is 25% greater than 4%.  Yet in the one group, 96% of people did not have a stroke, and in the other, 95% did not have a stroke.   

According to the abstract, they recorded 1310 cases of cerebral infarction among the entire population, which means that cerebral infarction occurred in 78% of all stroke cases (i.e. it was the main type of stroke in this population).  Thus, we know that the absolute incidence of cerebral infarction in the low meat group was not more than 4%, and conclude that in the high meat group not more than 5% of subjects had a cerebral infarction type stroke--which means that more than 95% of women eating the so-called high meat diets (more than 3 ounces daily) did not have a stroke.  The absolute difference between the two groups was not more than 1%, but by using relative risk, the authors get to report it as a 22% increase in risk.  Creative accounting.



If you read "Why Most Published Research Findings Are False" you will learn that if a study finds a relative risk difference of less than 100%, the odds are that the findings occurred either by chance, poor study design, or investigator bias.  This study falls in that category, bad science--which is true of almost all epidemiological studies like this.

Now take a close look at the last sentence of the abstract of the journal article:

"Fresh (unprocessed) meat consumption was not associated with total stroke or with any stroke subtype."

What?  In the immediately previous sentence, they stated that red meat was associated with a 22% relative risk in cerebral infarction type stroke, but the last sentence says that fresh meat was not associated with total stroke or any subtype, which would include cerebral infarction.

I feel confused. How can "high" red meat consumption be associated with a 22% greater risk of stroke, and yet not associated at all with total stroke or any stroke subtype?  It seems that the trick must be in combining both fresh and processed red meat to get the positive association.

As you can see from the title of the Reuters report, the first claim got plastered on the headlines, but the last was ignored.   What's up with that?




This study didn't show that eating a diet rich in red meat causes strokes.  On the contrary, it showed that at least 95% of people who eat the so-called high meat diet don't suffer strokes.  It also showed that if you massage the data correctly, you get get a result that will get media attention and support conventional preconceptions.

P.S.  If you like this post and want to see more like it, please consider making a small donation or a recurring subscription payment using the PayPal buttons in the right hand column.  Fighting fallacies is a full time job I love to do, but I need support to continue doing it.  Also consider sending a link to this post to all of your Facebook and other friends.