Seasonality of Pre-term Births

Matt found a conference paper that shows that the risk of pre-term birth is the lowest in spring, rising through summer and fall and the greatest in winter.
The paper, IMHO erroneously, focuses on the time of conception (because it is an easy marker used to calculate the supposed birth-date). Matt correctly shifts the discussion to the time of birth. After all, pre-term births are much more likely to be caused by something happening around that time than anything at the time of conception.
On the other hand, Matt, though cautiously and almost tongue-in-cheek, makes an attempt at an adaptive explanation. To do this, he had to consider not just the time of birth but also the duration of pregnancy which brings us back to the consideration of the time of fertilization.
I don’t think there is an adaptive explanation (Matt thinks so, too, but tries to come up with one anyway). Pre-term births are not healthy shorter pregnancies. They are pathological events. I don’t think that natural selection works on preserving pathological events.
Both in the primitive state, and in today’s modern society, winter is the most stressful time of year (remember that silly formula for January 20-something being the most stressful day of the year?). Spring is the best time of year: people are coming out of their Winter Blues (or even full-fledged SAD), there is an abundance of food and it is not as hot and scorched-earth yet. The timing of birth is determined by the fetus, not the mother, but the fetus will often respond to the perceived stress of the mother.
Out of the left field: Perhaps the increase of pre-term births in industrial societies (compared to developing world and/or the past times) has something to do with Christmas and its commercialization!