While hte Vitamin A and bone story has been around for a LONG time, this paper was a bit of a dissapointment.
What it did is get a questionairre and serum sample from 2841 male subjects in 1970-73, then on 1860 in 80-83 and finally on 1221 in 1990-93. 1990-93 was the only time that dietary data was collected. Note that it would have only been on ~60% of the baseline sample. All follow up on subjects was done through hospital records, so that it relatively good, with an independant registrar to diagnose the fractures.
For their stats, they show information based on the whole sample they could follow up. All the information of serum retinol (vit a) came from baseline...
The people in the highest quintile of serum retinol, had the only significant increase in fracture rate. And the 95% Confidence is relatively high, especially for the hip fracture rate (range 1.15-5.28) which shows there isnt enough subjects to get it closer. They didnt adjust for dietary intake (can influence fracture rate), and physical activity in questionaire, especially at baseline when subjects are 50 doesnt really show anything. Remember peak bone mass is gained in the early to mid 20's.
It does add to the body of data though, but doesnt mean much to weight training folks.