Infant Nutrition: Vitamin D and Gluten
I just read an article by Reed Mangels, PhD, RD, Nutrition for Young Vegetarians: Birth to One Year, from the Vegetarian Nutrition Dietetic Practice Group’s newsletter (Vegetarian Nutrition Update Vol XXI, No. 4, 2013). It is not available to the public, so I cannot link to it. But it inspired me to add some info on infants and vitamin D to the article Calcium and Vitamin D:
“A 2010 study found that breast milk was not a sufficient source of vitamin D (1). A 1985 study recommended exposing babies to 30 minutes of sun a week wearing only a diaper in order to provide sufficient vitamin D (2), however, the American Academy of Pediatrics recommends no sun exposure for infants. The authors of the 2010 study recommend that all infants get the RDA for vitamin D of 400 IU via infant formula or vitamin D drops.”
Mangels’ article also pointed out a recent study (3) that found, to quote the abstract, “the risk of [celiac disease] was significantly reduced in infants who were breast feeding at the time of gluten introduction (pooled odds ratio 0.48, 95% CI 0.40 to 0.59) compared with infants who were not breast feeding during this period.”
References
1. Liang L, Chantry C, Styne DM, Stephensen CB. Prevalence and risk factors for vitamin D deficiency among healthy infants and young children in Sacramento, California. Eur J Pediatr. 2010 Nov;169(11):1337-44. | link
2. Specker BL, Valanis B, Hertzberg V, Edwards N, Tsang RC. Sunshine exposure and serum 25-hydroxyvitamin D concentrations in exclusively breast-fed infants. J Pediatr. 1985 Sep;107(3):372-6. (Abstract only.) | link
3. Akobeng AK, Ramanan AV, Buchan I, Heller RF. Effect of breast feeding on risk of coeliac disease: a systematic review and meta-analysis of observational studies. Arch Dis Child. 2006 Jan;91(1):39-43. (Abstract only) | link
June 29th, 2013 at 3:21 am
“the risk of [celiac disease] was significantly reduced in infants who were breast feeding at the time of gluten introduction (pooled odds ratio 0.48, 95% CI 0.40 to 0.59) compared with infants who were not breast feeding during this period.”
http://imr.sagepub.com/content/28/5/234.refs
What if the mother doesn’t eat gluten and feeds her baby gluten laden foods?
June 29th, 2013 at 6:33 pm
Andreas,
I don’t think the study assessed this. It does say, “The actual mechanism through which breast milk protects against the development of CD is unclear. It could be that continuing breast feeding at the time of weaning limits the amount of gluten that the child receives, thereby decreasing the chances of the child developing symptoms of CD. Ivarsson et al found that children with CD received larger initial amounts of flour compared to controls. Breast milk is known to significantly protect against a number of infections including gastroenteritis. Infections of the gastrointestinal tract in early life could lead to increased permeability of the intestinal mucosa, allowing the passage of gluten into the lamina propria. Gut infections are also known to increase tissue transglutaminase expression and this could favour the generation of deamidated gluten peptides, triggering CD in susceptible individuals.” And they go on to mention IgA and T-cells.
The paper is free if you follow the link I included in the References.
June 29th, 2013 at 4:58 am
Jack,
I wanted to ask your opinion on something. I was discussing PUFAs with a scientist at the Harvard School of Public Health, and he mentioned that men should try to limit their intake of ALA (alpha-linolenic acid) because literature has shown an association between ALA intake and the risk of prostate cancer. When I went to Medline to see if the literature supports his statement, I found a wealth of supportive articles. Just to cite one of these:
http://jnci.oxfordjournals.org/content/86/4/281.abstract
I know you are cautious about ALA because of macular degeneration, but the above, at least in men with intact prostates, should suggest caution about ALA intake vis a vis the risk of prostate cancer. Do you agree?
On a personal note, I consume a tablespoon of mixed sprouted flaxseed/chia seeds and a number of walnuts per day, as well as two avocados, tofu and lupins. According to peacounter, I am consuming 4,663 mg of ALA per day. Is this too much? I have a strong family history of prostate cancer (father and several second-degree relatives – uncles and male cousins).
Thank you very much for your comments.
June 29th, 2013 at 6:41 pm
Dan,
I address the question of prostate cancer and ALA here:
http://veganhealth.org/articles/omega3#Prostate
I conclude that in amounts of 2 g per day, there’s nothing to worry about. I have a hard time believing that ALA causes prostate cancer at any amount, but that’s just my own sense and I certainly don’t have proof of that. As you like to point out, there are usually a whole lot of confounders in these studies. If I had a history of prostate cancer in my family would I be worried about 4.6 g of ALA per day? I don’t think I would be.
Please don’t take this as medical advice — as you know, I’m not a doctor.
June 30th, 2013 at 1:31 am
I should have been more clear with my question.
Would a mother still produce antigliadin-IgA antibodies if she doesn’t eat gluten foods but her child is fed gluten foods? I don’t believe so.
Presence of high levels of non-degraded gliadin in breast milk from healthy mothers.
http://www.ncbi.nlm.nih.gov/pubmed/9867098
Even though celiac disease is reduced in infants, there is research which indicates that some people, including myself, absorb other gliadin proteins without damage to the intestinal mucosa and experience health problems elsewhere throughout the body. Acne on my arms and slight morning joint inflammation for myself.
50 Shades of Gluten (Intolerance)
http://www.huffingtonpost.com/chris-kresser/gluten-intolerance_b_2964812.html
July 1st, 2013 at 3:32 pm
> Would a mother still produce antigliadin-IgA antibodies if she doesn’t eat gluten foods but her child is fed gluten foods? I don’t believe so.
You’re probably right.
June 30th, 2013 at 5:59 am
Jack, you’ve put my mind at rest a bit.
I was reading on another website about the Bradford Hill criteria for causality in epidemiological research (there are nine such criteria). The prime one is the “strength of the association”, which was taken to mean a risk ratio in excess of 2.0 (for example, smoking increases the risk of lung cancer by a risk ratio of 10 to 30 times, depending on the amount smoked). To invert that risk ratio, an RR/HR/OR of 2.0 would mean a protective risk ratio under 0.50. Few studies in nutrition support that range of risk ratios.
However, in the more recent ALA study below in Jamaicans, high ALA intake was still harmful in terms of prostate cancer risk:
Cancer Causes Control. 2012 Jan;23(1):23-33. Associations of whole-blood fatty acids and dietary intakes with prostate cancer in Jamaica.
Jackson MD, Walker SP, Simpson-Smith CM, Lindsay CM, Smith G, McFarlane-Anderson N, Bennett FI, Coard KC, Aiken WD, Tulloch T, Paul TJ, Wan RL.
Source: Department of Community Health and Psychiatry, University of the West Indies, Mona, Kingston, Jamaica.
OBJECTIVE:
To investigate the association of whole-blood fatty acids and reported intakes of fats with risk of prostate cancer (PCa).
DESIGN:
Case-control study of 209 men 40-80 years old with newly diagnosed, histologically confirmed prostate cancer and 226 cancer-free men attending the same urology clinics. Whole-blood fatty acid composition (mol%) was measured by gas chromatography and diet assessed by food frequency questionnaire.
RESULTS:
High whole-blood oleic acid composition (tertile 3 vs. tertile 1: OR, 0.37; CI, 0.14-0.0.98) and moderate palmitic acid proportions (tertile 2: OR, 0.29; CI, 0.12-0.70) (tertile 3: OR, 0.53; CI, 0.19-1.54) were inversely related to risk of PCa, whereas men with high linolenic acid proportions were at increased likelihood of PCa (tertile 3 vs. tertile 1: OR, 2.06; 1.29-3.27). Blood myristic, stearic and palmitoleic acids were not associated with PCa. Higher intakes of dietary MUFA were inversely related to prostate cancer (tertile 3 vs. tertile 1: OR, 0.39; CI 0.16-0.92). The principal source of dietary MUFA was avocado intake. Dietary intakes of other fats were not associated with PCa.
CONCLUSIONS:
Whole-blood and dietary MUFA reduced the risk of prostate cancer. The association may be related to avocado intakes. High blood linolenic acid was directly related to prostate cancer. These associations warrant further investigation.
See the key sentence, in my view, stated above: “whereas men with high linolenic acid proportions were at increased likelihood of PCa (tertile 3 vs. tertile 1: OR, 2.06; 1.29-3.27).”
Still I agree with the potential confounding issue, especially in case-control studies (n=435 in this one).
I think I will back off on the chia/flaxseed, as I seem to be getting enough ALA from all other sources (e.g. walnuts, avocados, lupin beans, tofu/soy).
July 1st, 2013 at 3:44 pm
Dan,
> The prime one is the “strength of the association”, which was taken to mean a risk ratio in excess of 2.0 (for example, smoking increases the risk of lung cancer by a risk ratio of 10 to 30 times, depending on the amount smoked). To invert that risk ratio, an RR/HR/OR of 2.0 would mean a protective risk ratio under 0.50. Few studies in nutrition support that range of risk ratios.
Very interesting. Seems like it would also require a relatively tight confidence interval.
And what about cases like ALA and prostate cancer where there are many findings, many of them not significant?
July 3rd, 2013 at 6:24 am
Re the comment “I was reading on another website about the Bradford Hill criteria for causality…”
There is no such thing as criteria for causation, and this has been the established agreement among expert epidemiologists for a long time. Anyone who still claims otherwise (e.g., that website the commenter was reading) is mostly just demonstrating that they do not know what they are talking about. (This is especially true if they say that there are nine, since at least one of the items on Hill’s list has long-since been shown to not be useful to consider.) Hill called his list “considerations”, and they can be reasonably called that — some things to think about when you are trying to assess whether an observed association is causal. Looking at the list is useful for those who have no clue as to what to think about, though no one who is expert in the field needs to do so.
The strength of the association is quite helpful in ruling out random error as an explanation for the entire association (i.e., that the association is just in the data and not an association in reality), though has no particular advantage over other considerations beyond that. It is not “prime” and there is most certainly no specific relative risk level below which an observation should be dismissed. Again, there are no “criteria” and there are certainly not any bright lines.
There is no particular reason to believe that smaller associations — if there really is an association — are much less likely to be causal than larger ones, and just throwing them out eliminates a lot of knowledge. Imposing a RR limit of doubling the risk would eliminate almost all knowledge about risk factors for heart attack, for example.
July 3rd, 2013 at 8:24 am
I see a distinct contradiction in these two sentences:
“There is no such thing as criteria for causation …. Anyone who still claims otherwise (e.g., that website the commenter was reading) is mostly just demonstrating that they do not know what they are talking about. ”
and
“they [Bradford Hill criteria] can be reasonably called …. some things to think about when you are trying to assess whether an observed association is causal.”
I would posit the question – What is the distinction between criteria for causality and “some things to think about [to] assess when an observed association is causal.”?
Since so few studies in nutrition posit a risk ratio above 2.0 (harmful) or below 0.5 (protective), the few that do tend to really stand out; hence, the importance of condition 1 (strength of association).
For example, having a BMI greater than 30 (a.k.a. being obese) has a risk ratio of more than 40 for subsequent onset of type 2 diabetes (>4000% increase in risk). Causality is likely here, even in the absence of the other criteria (biological plausibility, dose-response gradient, specificity, experimentation, etc), because the risk ratio is so extreme.
Conversely, when I see an observational study suggesting that intake of some dietary item is related to a relative risk increase of 11% (RR 1.11), I would tend to dismiss it. Even if causal, such a risk increase is probably clinically irrelevant in real world individuals (though depending on the degree of exposure and the size of the population, the impact measured by PAR could still be sizable, if it’s truly causal).
Saying things like “[they are] mostly just demonstrating that they do not know what they are talking about” is an ad hominem attack and does not advance the discussion helpfully.
July 3rd, 2013 at 2:17 pm
Um, no — no contradiction. A criterion is something that can be met (or not). There are no criteria for determining causation, as much as people might want a shortcut to avoid having to think hard about scientific inference.
It is true that a huge portion of nutritional epidemiology is junk science, but that is not because the true associations that we would like to identify are small. The reasons are more complicated. It is obviously the case that there are many low-RR true causal relationships, some of which are worth knowing and acting on. Just declaring that we will never draw any conclusions about these because of some (nonexistent) rule of thumb is rather obviously a bad idea.
As for your rule of thumb, positing your claim about the RR for obesity, it is presumably also the case that wearing size 42 jeans has an RR of about 40 for T2DM. It is also the case that wearing XXL t-shirts has an RR of about 40. I have now identified two very high RRs you could observe in the data (as compared to your one) that are not causal relationships. Your simple rule of thumb is losing already, and I could obviously come up with more examples.
Of course, a bit of reasoning (or, for a confounding problem — which this is — that is more complicated, some additional quantitative analysis) can lead us to the conclusion that the observed association is not causal. That is the point of the considerations that one might think about before concluding causation.
Finally, you might want to look up the meaning of “ad hominem”. My criticism was, roughly speaking, the exact opposition of that.