Google+ Followers

jeudi 28 janvier 2016

The debate around vitamin D

"RCT's not Designed for Public Health

How do we get information to make decisions about public health from randomized controlled trials (RCT's)? Can we? Especially for nutrient trials?

At this point, we have seen so many problems with the design of RCT's that get in the way of making public health decisions, i.e., does the nutrient benefit health? At what levels? What can we do?

Dr. Robert Heaney, our Research Director, has published a paper on the appropriate design of nutrient trials and, most of what we see is definitely not designed properly.

In the trial described below, The Vitamin D Antenatal Asthma Reduction Trial (VDAART), there are several areas where the information we need is missing:
  • They measured the effect of treatment by dosage treatment group, not by serum level. This will be an ongoing problem with any vitamin D trial. As we have published, there is a 6 fold spread of serum levels at any given dose: for someone taking 2000 IU/day, their level might be anything from 20 ng/ml to 120 ng/ml. A study that only analyzes doses will definitely have weak and incomplete results.
  • The lack of accounting for those who were not compliant, even with the dosage model, greatly confuses the interpretation of the data. When you are supposed to have 10 people and only have 7, the average is the sum divided by 10 vs. the sum divided by 7. A great mistake.
  • The lack of recognition of the public health impact vs. arbitrary statistical significance. If there is a P-value of 0.05, this means that there is a 5% probability that the results were due to chance, or, that you have a 95% probability that it was NOT chance. With the P-value of 0.051, this means that there is a 5.1% probability that the results were due to chance. When you are making a decision for public health, does 0.1% 'matter'? The focus on this arbitrary cutoff point distracts from the important public health findings in this study. Additionally, if a product is safe and inexpensive, like vitamin D supplements, then a higher P-value should be acceptable. There would be no harm caused and it is highly likely that it would help (with only a 5.1% chance that you are wasting your money).
Please note the detail below. We think this analysis was very misleading due to the requirements put on the RCT's, not on the researchers.
Thanks especially to you, our concerned audience, for your attention to health! We always appreciate your attention, your input.
Carole Baggerly 
Director, GrassrootsHealth 
A Public Health Promotion & Research Organization 
Moving Research into Practice NOW!"

Another piece:
"Does vitamin D during pregnancy reduce asthma in early childhood?

Wheezing illnesses begin as early as the first few weeks of life, suggesting that there may be prenatal determinants involved with the genesis of these illnesses. While wheezing and other asthma-like symptoms are common in the preschool ages, it remains difficult to predict which of these preschoolers will have asthma that persists through mid-childhood and onward. 
Dr. Bruce Hollis
Dr. Bruce Hollis, PhD, is one of the researchers responsible for the research behind our Protect our Children NOW! campaign. He has been involved in vitamin D research for 35 years. He is passionate about improving health during pregnancy and early childhood through vitamin D.

An Editorial Says New Study Exhibits "Inconclusive Results"

Hollis was an investigator for The Vitamin D Antenatal Asthma Reduction Trial (VDAART) which tried to determine whether prenatal vitamin D supplementation could prevent asthma or recurrent wheeze in early childhood. Hollis expected about a 25% reduction in asthma with vitamin D sufficient women (> 30 ng/ml). This randomized controlled trial included 881 pregnant women from 14 weeks gestation. The treatment group received 4400 vitamin D daily, while the control group took a placebo and a prenatal vitamin (400 IU/day of vitamin D). When analyzing by treatment group assignment, the incidence of asthma and recurrent wheezing in children at age 3 was lower by 6.1% among children from the treatment group compared to the control group (24.3% v. 30.4%). This data was said to have a P-value of .051, which is .001 higher than the general definition of "statistical significance." Because of this, JAMA published that the study was "not statistically significant."

"Not statistically significant"?

Let's address the P-value first. The P-value is an index for the strength of the evidence. For this study, a P-value of 0.051 would mean that if you did the exact study again with another group of 881 women, 51 of 1000 of these studies would have found a 6% or greater reduction due to chance. In a different wording, that means you are 94.9% sure it was NOT chance! For some arbitrary reason, in contemporary experimental studies, "statistical significance" is now conventionally set at a P-value of <0 .05.="" 0.1="" 4.9="" 5="" 95="" a="" above="" are="" as="" chance.="" chance="" decision="" div="" does="" due="" for="" have="" health="" is="" it="" making="" matter="" means="" nbsp="" not="" or="" probability="" public="" results="" stated="" that="" the="" there="" this="" to="" was="" were="" when="" you="">
6% Reduction

Next, let's address the 6% reduction. This might not be enough to grab headlines, or to make a change in the standard of care for pregnant women. Should we expect only a 6% reduction if all the women actually were taking the vitamin D? And, gave their children vitamin D? Hollis thinks it would be bigger. There were two things that could account for the lower percentage of reduction. 

The first is conforming to the daily dose. Compliance with the study medication was approximately 70%, as measured by Medication Event Monitoring System (MEMS). For each participant in the trial, their container of vitamins/placebo recorded the time and date of each opening and closing through integrated micro-circuitry. Unfortunately, non-compliance dilutes any true effect for intent-to-treat analyses, but patients could not be removed from the trial for non-compliance. This non-compliance reduced the vitamin D levels of the treatment group. While the levels did rise, they were not in the recommended range (100 - 150 nmol/L, 40-60 ng/ml ) - but instead averaged 98 nmol/L (39 ng/ml). Hollis speculates that 4,400 IU/day might not be adequate for all pregnant women. 

Next, there is the factor of the children's levels after birth. After birth only 46% were receiving vitamin D from their mother and this was only during the first year of life. The results show that the differences in asthma were significant only within a year of birth. This suggests the early effects of vitamin D through pregnancy do not last for three years.

What about their vitamin D levels?

Of course, the analysis we want to see is the comparison of the incidence of asthma according to the vitamin D levels of the mothers (and infants). When this analysis is completed, we predict that those mothers (and infants) who achieved and maintained 40 ng/ml or higher during pregnancy and the first three years of life would indeed have reduced overall asthma rates.

As this study hits the presses, please ask yourself the question about how to interpret this for public health (not a statistical audience): "If the probability is that there is a 94.9% chance that there is at least a 6% reduction in asthma, shall I see that the prenatal person gets vitamin D? Is it safe (yes), is it inexpensive/available?" From there, please make your best public health choice. Take a look at our disease incidence prevention chart for pregnancy."

Aucun commentaire: