Example of a vitamin D meta-analysis which ignored the data – June 2012

We never cease to be amazed at the stupidity of some meta-analysis

Sometimes they take no account of the IUs of vitamin D taken by each of the trials they are analyzing

Example: Vitamin D and gestational diabetes: a systematic review and meta-analysis
European Journal of Internal Medicine 2012

The authors ignore that there was no statistical difference in vitamin D levels in three trials
Farrant, Makgoba, and Baker stated that they had NO statistical difference between the two arms

YET, this meta-analysis included the data from those trials

Image

If we ignore the data from those 3 trials it appears that the odd ratio would increase to about 2.4 from 1.6

Image

If the authors had actually gotten around to consider the vitamin D levels actually measured by each of the trials
and weighted the data based of each based on the vitamin D levels (rather than no weighting at all)
the odds ratio would most likely had been far higher - perhaps >3

Farrant study is attached at the bottom of this page

Notes:

  1. The vitamin D levels were so low that they did not detect any difference in Gestational Diabetes
  2. The vitamin D levels were so low that they did not detect any difference in neonate size
  3. They compared ALL mothers to those mothers less than 20 nanograms,
    rather than the conventional comparison of those above to those below some level, say 20 nanograms
  4. Something is strange with the Farrant study.
    The women taking vitamin D supplements had lower level blood levels of vitamin D
    Wonder if the mothers were taking vitamin D2 instead of vitamin D3

See also VitaminDWiki

6106 visitors, last modified 23 Jun, 2012,
Printer Friendly Follow this page for updates