We have had a lot of emails about a recent study by Dror et al (1). They found a nonlinear risk of 25(OH)D levels, with levels above 36 ng/ml conferring a small but significant risk but levels below 20 ng/ml conferring a major risk. The authors measured 25(OH)D levels in more than 400,000 subjects and then followed them to see who had died or had heart attacks. After up to 54 months of observation, more than 3,900 subjects had died or had heart attacks. Compared to those with 25(OH)D levels between 20-36 ng/ml, the adjusted hazard ratios for heart attacks among those with levels of <10 ng/ml was 1.88 while those with 25(OH)D levels above 36 ng/ml, it was 1.13.
Most vitamin D experts believe the elevated risk in those with levels above 36 ng/ml in Dror et al were due to the fact that sick people tend to take or get vitamin D prescribed to them more often than well people do, thus skewing the curve. That is, more very sick people are taking or given vitamin D and that vitamin D comes too late to prevent them from dying.
Doctor John A. Eisman of New South Wales University in Australia wrote that Dror et al reported the attributable risk for mortality and acute coronary syndrome events is 32% for 25-OH D values below 20 ng/ml but only 0.55% for 25-OH D values above 36 ng/ml (2). Thus, if one were to take these risks at face value, it seems that the ratio of benefit to risk of indiscriminant vitamin D supplementation in an unscreened population would exceed 50:1.
Dr. Eisman also pointed out there was no evidence of a dose–response relationship at the higher levels of 25-OH D with virtually the same apparent adjusted risk for levels of 36–40 ng/ml, 40–44 ng/ml, and > 44 ng/ml. If higher blood levels of vitamin D were dangerous, then the higher the level the greater the danger but Dror et al did not find that. However, fewer than 1% of the subject had 25(OH)D levels greater than 36 ng/ml so further analysis of that group was problematic.
However I believe that nature knows best. I found Dror et al unconvincing because they claim they know more than nature and natural 25(OH)D levels are dangerous. I think this highly unlikely. By natural I mean those levels obtained by free-living hunter-gatherers in equatorial Africa, which average about 46 ng/ml. (3) I also do not know of a mechanism by which 25(OH)D levels above 36 ng/ml could be dangerous. Vitamin D toxicity may start with levels somewhere above 150 ng/ml but certainly not at 36 ng/ml. (4)
I theorize that Dror et al’s findings were due to confounding with vitamin A. In a Cochrane Review meta-analysis of randomized controlled trials, they found people who take vitamin A have a 16% increased rate of dying (5). Examining dietary intakes in the Nurses’ Health Study, Oh et al found that women in the highest quintile of vitamin D intake also took the most vitamin A, about 10,000 IU/d of retinol, much higher than what anybody recommends (6).
Cheng and Neuhouser have recently shown that a very significant nutrient-nutrient interaction occurs between vitamin A and vitamin D (7). They found that elevated retinyl ester levels were common among those with high 25(OH)D levels and those retinyl ester levels confounded 25(OH)D’s favorable effect on lung cancer. Excess vitamin A intake and subclinical vitamin A toxicity is a real problem (8). Mice fed excess vitamin A develop heart disease (9). Vitamin A also increases CRP (10).
So we know that excess vitamin A causes death, that people who have the highest vitamin D intakes also have highest vitamin A intakes and that a significant nutrient-nutrient interaction occurs between vitamin A and vitamin D. If Dror et al had measured serum retinyl esters in his population, current evidence predicts he would have found elevated retinyl esters in those with the highest 25(OH)D levels, thus confounding their study.
Are levels above 40 ng/ml a good idea? In a recent study of gene expression, the person in the study who had an initial 25(OH)D level of 30 ng/ml, and who obtained 25(OH)D of above 40 ng/ml after supplementation, had differential expression of 33 genes, suggesting there is a significant benefit of 40 ng/ml (11). It has been shown that bisphosphonates work best when the 25 hydroxy vitamin D level are greater than 40 ng/ml (12).
Bone mineral density progressively increases as the serum level of 25 hydroxy vitamin D increases to 40 ng/ml or more (13). There are additional benefits in maintaining 25(OH)D levels even higher than 40 ng/ml, for example autoimmune disease, cancer and diabetes (14). Pregnant women supplemented with 4,000 and 6,400 IU/day had fewer complications of pregnancy than did women on 400 IU/day and many of those women had levels above 36 ng/ml (15).
However, as a zoologist, I go back to the common sense idea that nature knows best. Mean natural 25(OH)D levels are around 46 ng/ml. In fact, a follow up study showed that pregnant hunter-gatherers in Tanzania had mean levels of 60 ng/ml and one pregnant hunter-gatherer was above 100 ng/ml (16). Until a randomized controlled trial shows high dose supplementation is dangerous, I will maintain natural vitamin D levels.