This was not simply observational. There will be no correlation between being given higher dose vitamin D and age in this trial, the results appear at first blush to be showing Vitamin D is working independant of age.
By the by this /exactly/ why a randomised trial has been the gold standard for medical treatment for so long. False correlations, confounding factors etc are all vastly less likely to skew the results. 2 groups chosen at random, nobody in the groups or treating them knows which is which. 1 group given treatment, 1 given a placebo. How much difference do we see in the 2 groups as a result of treatment.
It's a good question. It's always worth asking. It's always worth checking. The rabbit hole of stat analysis of treatments goes pretty deep from there. Ethical issues come in. Expense. But we all need to make the effort to understand it on some level or we're marks for snake-oil.