Posted by: Christopher Masterjohn in WAPF Blog on 03/09/10
Many people have asked me or the Weston A. Price Foundation to respond to a recent newsletter put out by Dr. John Cannell, MD, of The Vitamin D Council claiming that cod liver oil and vitamin A supplements are “poison” and that our country’s “problem is widespread subclinical vitamin A toxicity.” The Mother Nature Obeyed blog is the perfect place to respond.
I would like to start out by saying that I respect Dr. Cannell’s efforts to raise awareness about the need for greater exposure to sunshine and vitamin D supplementation when adequate sunshine is unavailable, and that this post is not in any way a personal attack against Dr. Cannell or his work. Disagreement in science is the stone on which the knife of truth is sharpened, so in that sense Dr. Cannell and I are partners in cutting through the obscurity, uncertainty, and falsehood that lies in the way of achieving robust and radiant health.
Researchers are increasingly examining the interactions between vitamins A and D. The most recent study to do this was a January, 2010 study published in the British Medical Journal looking at the relationship between blood levels of vitamin D and the risk of colorectal cancer. The study was nested within the massive EPIC study, containing over a half a million subjects, but the vitamin D study itself contained just under 2,500 subjects. They found 1248 people within the original, large study who developed colorectal cancer after enrollment and matched them to an equal number of people from the same study who did not develop colorectal cancer during the same period of time. Then they compared dietary information and blood levels of vitamin D between the two groups, taken back when the subjects were first enrolled.
Before we consider the results of the study, let’s review a few basic principles of the scientific method. When we approach science, we first observe the world around us. When we come across an interesting observation in need of an explanation, we formulate a testable hypothesis as a potential explanation. We then perform an experiment to test the hypothesis. The results of the experiment allow us to confirm, discard, or revise the hypothesis.
Studies published in nutrition and medical journals generally fall into one of two categories: observational and experimental studies. Observational studies in these cases are usually looking at existing statistical relationships between different variables. In our particular case, we are looking at the statistical relationship between blood levels of vitamin D and the risk of colon cancer, so this is an observational study. Experimental studies also use statistics in most cases, but their key feature is that the investigators involved perform some intervention in humans, animals, isolated cells, or purified biological molecules. They will randomly distribute the study subjects or study materials into two or more groups, at least one of which is not treated with the intervention and therefore serves as the control.
According to the scientific method, the first category of studies is useful for generating hypotheses, but not for determining whether those hypotheses are true. The second group of studies, by contrast, allows us to confirm, discard, or revise the hypotheses generated by the first group.
Most modern studies test the effect of an intervention on a statistical endpoint. The researchers aim to use the treatment to increase or decrease the likelihood of a particular result. According to the laws of statistics, we can only infer whether one thing causes another if we randomly distribute study subjects into treatment and control groups. If we performed an experiment with vitamin D supplementation and then allowed people to pick whether they wanted to be in the vitamin D group or the control group, a statistician would tell us we would have to treat the study as an observation instead of a true experiment. If we found that the vitamin D group had a lower risk of cancer, the statistician would tell us that we had no basis for claiming that the vitamin D caused the lower risk of cancer. After all, in this hypothetical example, what caused some people to want to take vitamin D and others not to want to take it? We have no idea, but whatever those factors were could have also caused the reduction in cancer.
When we perform observational studies like the one published in the British Medical Journal, we try to make statistical adjustments for all of those unknown factors – we call them “confounding variables” – but the truth is that there will always be much more we don’t know about the world than we do know, so we never treat these studies as controlled experiments. After all, is there any way to statistically adjust for whatever psychological trait put someone in the mood to go out in the sun more often and go for walks, fly kites, go hiking, do cartwheels out in the park, go swimming, or do whatever else these people were doing outside in order to obtain higher vitamin D levels? Of course not.
So, back to the results of the study. Among those study subjects who developed colorectal cancer, there were more subjects with low vitamin D levels at the beginning of the study and fewer subjects with high vitamin D levels at the beginning of the study than there were among those who did not develop colorectal cancer. The authors used statistical calculations to estimate that subjects with 25(OH)D levels above 40 ng/mL were 23% less likely to develop colorectal cancer and those with levels between 20 and 30 ng/mL, and that those with levels below 10 ng/mL (extremely deficient) were 32 percent more likely to develop colorectal cancer.
So where does vitamin A come in? Vitamin D status seemed to be related to the risk of colorectal cancer only in people consuming very low amounts of vitamin A.
The researchers split people into three groups according to vitamin A intake: those who consumed less than 1500 IU/day, those who consumed more than 3000 IU/day, and those who consumed some amount in between those two values. To put this in perspective, the RDA for vitamin A is 3000 IU/day for men and 2300 IU/day for women. In those consuming less than 3000 IU of vitamin A per day, low vitamin D status was associated with an increased risk of cancer and high vitamin D status was associated with a decreased risk of cancer. In those consuming more than 3000 IU of vitamin A per day, however, the magnitude of these relationships became so small that they lost statistical significance, which means the effect of vitamin D status was so small that it could not be distinguished from the effect of chance.
Naturally, opponents of vitamin A supplementation like Dr. Cannell have seized on the fact that high vitamin D status was not associated with the benefit of a decreased risk of colorectal cancer in those consuming the RDA for vitamin A. They have, however, ignored the fact that low vitamin D status was not associated with the harm of an increased risk in the same population. And thus they claim without any true justification that vitamin A intakes at or above the RDA render vitamin D useless and that vitamin A-rich cod liver oil constitutes "poison."
Of course, it is a far leap from "vitamin A" to "cod liver oil" in this case because there is no indication where the vitamin A came from in this study. Perhaps some subjects took cod liver oil and others took multivitamins or drank fortified milk and ate fortified breakfast cereals. The vitamin A in natural foods, including cod liver oil, could act differently than the vitamin A added to fortified foods and to supplements.
It is most important, however, that we recognize the proper place of this and similar studies within the scientific method. As observational studies, they offer us the chance to formulate a hypothesis, but they do not “prove,” “show,” or “demonstrate” that the hypothesis is true.
The authors of the of BMJ study themselves recognized these limitations. “The strong inverse associations of the present observational study,” they wrote, “suggest that further research efforts should concentrate less on observational findings and more on clinically relevant studies to determine whether vitamin D has a causal role in colorectal cancer prevention or whether it is a marker of other events.” Similarly, noting that the relationship between vitamin D status and colorectal cancer was more statistically robust for 25(OH)D levels under 75 nmol/l (30 ng/mL), the authors wrote, “This finding suggests that raising very low levels of 25(OH)D to the mid-range may protect against colorectal cancer, and that levels of 75 nmol/l might not significantly reduce the cancer risks any further, but this needs to be proven in a clinical trial.”
Some people, Dr. Cannell for example, have looked at observational studies like this and offered the hypothesis that vitamin A antagonizes vitamin D and that most people in modern society consume too much of it. I have offered an alternative hypothesis. In my articles “From Seafood to Sunshine: A New Understanding of vitamin D Safety,” and “The Cod Liver Oil Debate,” I have argued that vitamins A and D act as molecular partners and that it is important to consume a diet rich in both vitamins in order to make each safe and effective. I also incorporated this view into my peer-reviewed hypothesis paper, “Vitamin D Toxicity Redefined: Vitamin K and the Molecular Mechanism,” a hypothesis that was partly confirmed in an animal study conducted by researchers at Tufts University, which I wrote about over at The Daily Lipid.
To settle these opposing hypotheses, someone needs to conduct a study where humans are given vitamin D supplements or a placebo on a vitamin A-restricted diet, combined with several levels of vitamin A supplementation. As discussed earlier, the researchers would have to randomly distribute the participants to the different treatment groups.
In the mean time, there is some uncertainty over this question. We can look at other observational studies, but we should realize when we do this that we are only lumping observation on top of observation. This might help us fine-tune our hypothesis but it does nothing to help us test it. When we get to the point where we are trying to estimate vitamin A intakes in studies where they are not recorded, we are lumping speculation on top of observation and it does even less to help us test the hypothesis.
In the absence of certainty, I prefer to follow the maxim after which this blog is named: that “life in its fullness is Mother Nature obeyed.” We know that traditional diets developed over thousands of years by groups all over the world that had successfully fine-tuned nutritional strategies that allowed them to pass radiant health from generation to generation consumed diets rich in foods like liver, other organ meats, cream and butter, shellfish, or insects, that provided a rich assembly of fat-soluble nutrients that included plenty of vitamin A. We will continue to learn more about nutrition as science progresses, but we should have a better foundation than a handful of unexplained statistical correlations on which to act in the face of uncertainty. In such cases, it makes the most sense to defer to the wisdom of traditional diets.
- - - - - - - - - - - - - - - -
- Vitamin D needs some vitamin A, but not too much
- All items in category Vitamin A
56 itemsResponse by Masterjohn on vitamin A thwarting vitamin D – Mar 2010 2683 visitors, last modified 26 Oct, 2012,This page is in the following categories (# of items in each category)