Posted by Larry Hoover on March 21, 2009, at 12:00:35
In reply to Re: Vitamin D May Not Be the Answer to SAD » Larry Hoover, posted by jrbecker76 on March 18, 2009, at 15:16:04
Thank you for sending me the .pdf file, monsieur becker. Much easier to interpret the data in their proper format.
I had a heck of a time trying to reconcile these two passages from the text, the first one appearing in the first paragraph of the "Results" section, and the second appearing in the first paragraph of the "Discussion".
"The prevalence of depressive symptoms was lower in the top tertile of 25(OH)D compared to the lowest tertile (78 vs. 121 cases of 1087 participants, or 7.2% vs. 11.1%) in the study population (odds ratio=0.62, 95% confidence interval=0.46 0.83, P=0.001, P for trend=0.002, Table 1). This association was substantially attenuated after controlling for aforementioned confounding factors, and disappeared after including geographic location in the model."
"The robust relation observed in the crude analysis was mainly due to the strong association of depressive symptoms and 25(OH) D with geographic location."
The more I looked at the statistics in Table 1, the less sense the stats seemed to make to me. Controlling for confounding variables should not have led to the derivative statistics for risk of depression under the various models which adjusted for confounds, based on my own understanding of the effects of the identified confounds. Then, finally, I had the answer. There is a huge error somewhere in the data!
They have more depressive cases reported in Beijing under the highest tertile of vitamin D levels (81 cases in 541 subjects) than they have reported for that same category in the whole study (78 cases among 1087 subjects in Beijing and Shanghai combined)! The percent incidence for third tertile depression in the whole study (7.2%) ought to be roughly the mean of the incidence for the two sites (15.0 and 3.7%), but it's not. The adjusted derivative stats appear to be consistent with the erroneous data reported for Beijing. There's no way to know what the correct data are, but the study should be dismissed as unreliable until they can sort out this discrepancy.
Lar
poster:Larry Hoover
thread:885916
URL: http://www.dr-bob.org/babble/20090313/msgs/886384.html