Spontaneous loss of very early pregnancies is very hard to study (because they may happen without knowledge, or at least confirmed conviction, that they were there to lose) so:
a) Apparent false positives of super-early tests could be falsely false, but then the actual pregnancy went wrong, hidden by other factors (conversely, false-negagatives might be hidden by the same mechanism),
b) Post-conception 'periods' might be mere (sic) bleeds, not always fatal to the carrying to term, even if it's a recognised danger sign.
c) The lack of a period (or two) isn't proof positive of pregnancy, as there may be other physiological reasons that spontaneously crop up. Although this could also compound with that possibility of very very early 'miscarriages' in some circumstances.
I really have no knowledge of the efficacy or utility of the current range of tests, no practical experience[1] of having a menstral cycle/possibility-of-pregnancy in a first-hand manner, and please don't take this as a claim of expertise in the minutiae. It's just a rounding up of some of the kinds of confounding factors that certainly make the 30 days limit pretty much beset by practical flaws that might provide for too many edge/over-edge conditions to make it an equitable solution even if you accept the premise of the earliest intervention being the least abhorent compromise.
(Really, once you get to that stage, you might as well go full-Romania. Which, luckily, they can't. Although I imagine there's also enough other pseudo-Romanian practices in effect if you investigate deep enough in the right places.)
[1] Given the acknowledged differences in experience, that'd be somewhat anecdotal but I lack even that concession and not sure I can cross-correlate what little second-hand 'bystander' information I have gathered through the fog of arms'-length non-study (and probably deliberately kept clouded from me).