This is barely above junk. Most strikingly, there is no data that shows trends prior to the introduction of the helmet law. Therefore there is no model for comparing the before and after scenario to the "do nothing scenario", which is what you care about.
There are good, well-known, well-understood methods for handling all the problems with this paper, but since the author hasn't used any of them this work does not do more than incrementally increase the plausibility of the proposition that "Helmet laws make cycling less safe".
There are good, well-known, well-understood methods for handling all the problems with this paper, but since the author hasn't used any of them this work does not do more than incrementally increase the plausibility of the proposition that "Helmet laws make cycling less safe".