Will they EVER learn statistics?
The paper Association of Animal and Plant Protein Intake With All-Cause and Cause-Specific Mortality, an “original investigation” (whatever that means, since they just crunched numbers), published on the 26th of August–this paper is behind a paywall. Interestingly, a nearly identical paper with a nearly identical title, published in the same journal in 2016: Association of Animal and Plant Protein Intake With All-Cause and Cause-Specific Mortality, with the same outcome. And what is that outcome?
Animal Protein Causes Earlier Death
Instead of writing about the article, I just provide the commentary I sent to JAMA Internal Medicine last night, and which published this morning. It explains everything. If you wish to read my commentary online, visit here and scroll down to comments–mine is the first one. Here it goes:
August 29, 2019
Concerns about Misleading Interpretation and Incorrect Conclusion
Angela Stanton, PhD | Stanton Migraine Protocol Inc.,
The article by Budhathoki et al.,(1) is akin to the article in 2016 by Song et al (2). A study of association is used to support causation without appropriate statistical proof.
The article states in the abstract that “Intake of animal protein showed no clear association with total or cause-specific mortality” and then in the next sentence “In contrast, intake of plant protein was associated with lower total mortality.” If there is no clear association between meat protein and mortality, how can plant protein be associated with lower mortality? If there is no association, then there is no association.
In my view, the authors main findings are incorrect, confusing, and misleading.
This research is based on food frequency surveys for the previous year, completed once every five years. Food frequency questionnaires are questionable at best, particularly when a past year’s consumption has to be recalled.
A frequent error in a food frequency analysis is the substitution of plant protein for animal protein “on paper” with “isocaloric substitution interpretation,” without the subjects actually changing their diet. Such substitution cannot be used to conclude what would have happened had they actually changed their diet. So we cannot infer if their mortality changed.
And finally, none of the hazard ratios shown meet the Bradford Criteria of 2 to suggest that the associations are significant enough to consider causation for even further analysis, let alone conclude any causal significance. This study shows no association of mortality with the type of protein consumed.
1 Budhathoki, S. et al. Association of Animal and Plant Protein Intake With All-Cause and Cause-Specific Mortality. JAMA Internal Medicine, doi:10.1001/jamainternmed.2019.2806 (2019).
2 Song, M., Fung, T. T., Hu, F. B. & et al. Association of animal and plant protein intake with all-cause and cause-specific mortality. JAMA Internal Medicine 176, 1453-1463, doi:10.1001/jamainternmed.2016.4182 (2016).
There. I think the trash quality of the article is clear.
Comments are welcomed, as always, and are monitored for appropriateness