Well as an intelligent and scientific mind you might be inclined to shout out “No, why should it?”. But what if I told you there is statistically significant evidence that this is the case and that this was published in a respectable clinical journal? Not so sure anymore?
Well then you have fallen into one of the most effective traps of statistics. Correlation does not imply causation (and indeed that is the point the paper cited above is making). There are many scenarios where in which two factors, A and B, might be correlated.
I) A might cause B
II) B might cause A
III) A and B might partially cause each other
IV) A and B might be caused by a common third factor C or A is caused by C which is correlated with A.
Confused yet? Let me give some examples:
A is correlated to B, B causes A: You can see a lot of people with umbrellas when it’s raining. Therefore umbrellas cause rain.
A and B cause each other: Students with test anxiety fail test more often.
A and B are caused by C: When ice-cream consumption increases, more people drown. Obviously (in this case) heat causes people to buy ice-cream and go swimming. This is for example far less obvious for the correlation of HDL levels (good cholesterol) and a lower chance to get a heart attack. However, increasing HDL levels by medication does not reduce the risk of heart attack.
So what about our astrological signs and illness? Do I, as a scorpion, have to worry about an abscess in the anal and rectal region (P= 0.0123)? I sure hope not! And indeed there is a fifth case in which correlation is purely coincidental.
So if correlation does not imply causation, what does? Statisticians are working ways to be more certain if two correlating factors are really connected, but we are still missing the tools to reach absolute certainty.
– David Huesmann
 P. C. Austin, M. M. Mamdani, D. N. Juurlink, J. E. Hux, J. Clin. Epidemiol. 2006, 59, 964–969.