Hey Nick,
Good to see you are still checking the boards. Putting on my epidemiologist hat, my first thought is to check the magnitude of effect and statistical significance. The odds ratio (i.e. the relative likelihood) of a negative split for those wearing Vaporfly shoes vs. those not wearing them was 2.72--a fairly high odds ratio--and the p-value was 0.051. Not half bad.
Next are the potential confounders, ie. factors that could be associated with both negative splits and wearing the Vaporfly shoes but that have no bearing on your actual hypothesis of interest. It seems that your biggest potential confounder here is general running savvy. Runners who are savvy might decide to give these new Vapor Fly shoes a try. They also might know how to pace themselves better. In your particular case, the effect of "savvy" can be tested. Look up the names of these runners in a database like athlinks and compare those who wore the Vaporfly in NY vs. those who did not. If Vaporfly wearers were not more likely to negative split in past marathons (those pre-dating sale of the Vaporfly), you can (mostly) rule out the "savvy" theory.
Finally, if you rule out "savvy", you or someone else would be justified in proceeding to a small controlled study, as suggested by others on the board. Given the magnitude of effect observed, it could be a small affordable study.
To wrap up, in my estimation, the results are definitely suggestive and will likely be followed up further by someone (if not necessarily you) given the magnitude of effect and the fact that it could be tested in a small(ish) study.
Best,
Steve