Thursday, October 17, 2013

More on the Harvard study on saturated versus polyunsaturated fats

This is a follow up on this publish, which tackled the main argument place forth in a latest BBC article. The BBC write-up argued that folks need to substitute saturated with polyunsaturated fat to lessen their risk of coronary heart condition.

Let us get a look at the true Harvard examine by itself (i.e., the review talked about in the BBC report). The Harvard examine is connected below.

This submit, by Stephan Guyenet, presently pointed out many troubles with the research. Stephan actually reviewed the reports utilised in the meta-investigation, and also some that ended up excluded in the meta-analysis and that he believes need to have been incorporated.

Here are a handful of other troubles, in addition to the types already pointed out by Stephan:

A single point that appears suspicious about this Harvard meta-analysis study is that they say that: “Statistical evidence for substantial between-examine heterogeneity was not existing (Q-statistic p = .13 I2 = 37%).”

A meta-investigation is a examine that essentially summarizes, in a statistically sophisticated way, a bunch of other studies (the “sourced” reports). Too a lot amongst-examine heterogeneity (i.e., widely disparate final results between sourced scientific studies) is undesirable, simply because it can bias the outcomes.

The issue is related to that of striving to summarize internet value figures (e.g., by calculating their average) in a middle class neighborhood that takes place to have a few billionaires dwelling in it. The heterogeneity in wealth could guide to a wildly overestimated common.

Now, we know that p values go down with sample dimensions, and are usually large with tiny samples unless of course the influence calculated by the statistic is really robust, regardless of the statistic used.

Nicely, with a sample of only eight scientific studies, their p worth (related with the Q statistic) is shut to getting substantial at the .05 degree!

If this sample of sourced studies have been a tiny greater (say, twenty), there would be considerable between-study heterogeneity, which would get in touch with the meta-evaluation into issue. This is a massive issue, considering that a excellent meta-evaluation is expected to contain a huge amount of studies (e.g., better than one hundred), and this 1 integrated only 8 studies.

In addition, to the greatest of my understanding, the Q statistic is not quite trustworthy when employed with small samples, owing to its minimal electricity as a test of heterogeneity. This helps make the p benefit reported even far more problematic.

Finally, the sourced research with the largest sample (n = nine,057 hence possibly the most credible), indicated as “Minnesota CS” on Determine 2 of the Harvard study, found enhanced danger of coronary heart condition connected with increased usage of polyunsaturated fats and diminished use of saturated fat.


Mozaffarian, D., Micha, R., & Wallace, S. (2010). Consequences on Coronary Coronary heart Disease of Growing Polyunsaturated Fat in Spot of Saturated Fat: A Systematic Overview and Meta-Analysis of Randomized Controlled Trials. PLoS Med., 7(three): e1000252. doi: 10.1371/journal.pmed.1000252.
Title: More on the Harvard study on saturated versus polyunsaturated fats
Rating: 910109 user reviews.
Posted by: Admin Updated at: 4:32 AM

No comments:

Post a Comment