Reader Comments
Post a new comment on this article
Post Your Discussion Comment
Please follow our guidelines for comments and review our competing interests policy. Comments that do not conform to our guidelines will be promptly removed and the user account disabled. The following must be avoided:
- Remarks that could be interpreted as allegations of misconduct
- Unsupported assertions or statements
- Inflammatory or insulting language
Thank You!
Thank you for taking the time to flag this posting; we review flagged postings on a regular basis.
closesame data-source, but almost exact opposite results
Posted by DrSiebler on 03 Mar 2014 at 16:53 GMT
The same data-source (Austrian Health Interview Survey AT-HIS 2006/07) was analyzed from the same authors, but resulting in the very antithesis of this study: http://link.springer.com/...
In the other study, published in February 2014, Nathalie T. Burkert and the other authors concluded:
"Our results show that a vegetarian diet is associated with a better health-related behavior, a lower BMI, and a higher SES. Subjects eating a carnivorous diet less rich in meat self-report poorer health, a higher number of chronic conditions, an enhanced vascular risk, as well as lower quality of life."
To my understanding of serious scientific, if someone get results A and !A regarding to the same data-source, then there seems to be something wrong with the methods. Under that conditions it seems very unclear (or even more than that to me), if the results can have a value at all.
Maybe the authors can comment on that? Why can one and the same data-source result in almost exact contrary results?
RE: same data-source, but almost exact opposite results
burkert replied to DrSiebler on 05 Mar 2014 at 10:49 GMT
Our different study results are due to the fact that in our publication in the “Wiener klinische Wochenschrift - The Central European Journal of Medicine” we analyzed the data of all 15474 persons which participated in the Austrian Health Interview Survey 2006/07 concerning differences between various forms of diet in health-related variables (comparing 2.2% vegetarians vs. 23.6% consuming a carnivorous diet rich in fruits and vegetables vs. 48.5% eating a carnivorous diet less rich in meat, and 25.7% consuming a carnivorous diet rich in meat).
In contrast, in the current publication in Plos One, we analyzed the data of 1320 subjects (who participated in the AT-HIS 2006/07) concerning health differences between subjects of the same sex, age, and socioeconomic status but consuming a different form of diet. The total number of analyzed subjects comprised 330 vegetarians (1) who were each matched to 330 subjects consuming a carnivorous diet rich in fruits and vegetables (2), 330 individuals eating a carnivorous diet less rich in meat (3), and 330 subjects consuming a carnivorous diet rich in meat (4). Each dietary habit group was set-up according to the same demographic characteristics concerning their age, sex, and socioeconomic status.
RE: RE: same data-source, but almost exact opposite results
DrSiebler replied to burkert on 06 Mar 2014 at 17:52 GMT
But nevertheless, in fact you stated two complete opposite thesis from one and the same data-source:
(1) "Our results show that a vegetarian diet is associated with a better health-related behavior, a lower BMI, and a higher SES. Subjects eating a carnivorous diet less rich in meat self-report poorer health, a higher number of chronic conditions, an enhanced vascular risk, as well as lower quality of life."
(2) "Moreover, our results showed that a vegetarian diet is associated with poorer health (higher incidences of cancer, allergies, and mental health disorders), a higher need for health care, and poorer quality of life."
In both studies you did make statements about "vegetarian diet" and "non-vegetarian diet". The statements are obvious complete opposite to each-other. So again: (non-)"vegetarian diet" can not be A and the !A (opposite-of-A) at the same time. It is logically impossible that both statements are valid at the same time if, as here, both assertions refer to one and the same object of study ("(non-)vegetarian-diet") and even to the same data-source.
I'm sorry, but if two methods result in thesis and the very antithesis using the same data-source: for me, there is no stronger evidence possible proving that the used methods must have serious deficiencies and so the conclusions about (non)-vegetarian-diets are rather worthless here (in both studies)..
RE: RE: RE: same data-source, but almost exact opposite results
burkert replied to DrSiebler on 11 Mar 2014 at 07:23 GMT
Actually, although the data basis was the same (data of the Austrian Health Interview Survey) we analyzed different data in both studies. The aim of the study published in Plos One was to analyze differences in terms of health indicators between individuals who are comparable according to their socio-demographic characteristics but differ in their dietary habits.
Our conclusions refer one time to eating behavior of the general Austrian population (Wiener Klinische Wochenschrift), and to eating behavior when certain socio-demographic characteristics are matched the same the other time (Plos One).
Since more women, younger persons, and individuals with a higher socioeconomic background consume a vegetarian diet the results of both studies are not comparable.
RE: RE: RE: RE: same data-source, but almost exact opposite results
MylesI replied to burkert on 24 Mar 2014 at 15:48 GMT
Then these results are devoid of any generalizable data. If your goal was to subtype your prior work, the conclusions on this paper should be “rich young female vegetarians may be sicker”. At best, this would suggest teen-to-30-something women are not the best at both avoiding meat and eating a balanced diet; instead we get comments about public health measures that are needed to prevent the harmful effects of eating vegetables. Additionally, if you wanted to ‘match’ your subjects, why have such a wild difference in the groups? Table 1 shows a range from 15yo poor boys to 80yo middle class females. With only 330 in each group, this is hardly ‘matched’ by ‘certain socio-demographic characteristics’ and not compared to the 15K subjects study reflecting the general population. If you wanted to subgroup, you should have done so with each tier presented in Table 1 – ie vegetarianism impact on 15-19yo, low income only, females only, etc. Or you should have age-adjusted, multi-varient adjusted your prior paper. Instead you simply lowered your power and reran your prior analysis.
I agree fully with DrSiebler that your papers are arguing the exact opposite thing. While I find your explaination somewhat acceptable, your discussion and abstract never bring up your prior work and never outline such nuance. Given the media’s love of jumping on hyperbolic nonsense, suggestions that ‘vegetarianism causes cancer’ are unethical if you are going to put them out there without any context. What would you want people to take away from both papers in total? Eat vegetarian or not? Limit meat or not? Only eat vegetarian if you are not a rich Australian women under 50? This is what I think you have not answered from DrSeibler – what “public health measures” are you advocating for in paper A and paper !A?
RE: RE: RE: same data-source, but almost exact opposite results
hgremmels replied to DrSiebler on 25 Mar 2014 at 00:22 GMT
My compliments to Dr. Siebler for finding the conflicting study by the same group. I find the discrepancy interesting, but not neccesarily damning for either study. For the primary research question - the effects of a vegetarian diet - the design of the Plos One study seems well suited. It's surprising, but not inconceivable, that the matched subset gives a different overall finding to compared multivariate model of the whole study population. With all parameter interactions in a multivariate model, I'm inclined to give more credence to the matched subset.
How was the matching done? I assume for the rare vegetarians, there must have been multiple (near-) matches, did the authors try the analysis with different sets of matched controls and were the results stable?
At any rate, a very interesting study; it's a pity the study in the Wochenschrift was published near simultaneously, which I assume is the reason the authors don't mention the results in the discussion. Perhaps the conficting results will finally motivate somebody to do a proper prospective study in the future.
RE: RE: RE: RE: same data-source, but almost exact opposite results
burkert replied to hgremmels on 26 Mar 2014 at 15:28 GMT
In our opinion, it seems not far to seek that persons with worse health consume a vegetarian diet because they try to develop a better health and eating behavior, and not the opposite, that a certain diet (vegetarian) leads to worse health. We therefore state in our discussion that we can neither say anything about causes or effects, nor about long-term consequences. Moreover, we say that further studies are needed to analyze nutritional habits and their association with health.